Skip to main content

Full text of "Sputtering of chemisorbed nitrogen from the (100)--planes of tungsten and molybdenum: a comparison of computer simulation and experimental results."

See other formats


AL POSTGRADUATE SCHOOL 
MONTEREY, CALIFORNIA 93943-5008 



NAVAL POSTGRADUATE SCHOOL 

Monterey, California 








THESIS 




SPUTTERING OF CHEMISORBED NITROGEN FROM THE (100) 
PLANES OF TUNGSTEN AND MOLYBDENUM: A COMPARISON 
OF COMPUTER SIMULATION AND EXPERIMENTAL RESULTS 






by 








Philip Jay Mattson 








December 1986 




Thesis 


Adv 


isor: Don E. Harrison, Jr. 





Approved for public release; distribution is unlimited 



T 2348? 5 



UNCLASSIFIED 

SECuSirv Classification Of This page 



REPORT DOCUMENTATION PAGE 



la REPORT SECURITY CLASSIFICATION 

Unclassified 



1b RESTRICTIVE MARKINGS 



2a SECURITY CLASSIFICATION AUTHORITY 



2b DECLASSIFICATION /DOWNGRADING SCHEDULE 



3 DISTRIBUTION /AVAILABILITY OF REPORT 

Approved for public release; distribution 
is unlimited 



4 PERFORMING ORGANIZATION REPORT NUMBER(S) 



S MONITORING ORGANIZATION REPORT NUMBER(S) 



6a NAME OF PERFORMING ORGANIZATION 

Naval Postgraduate School 



6b OFFICE SYMBOL 
(If applicable) 

Code 61 



7a NAME OF MONITORING ORGANIZATION 

Naval Postgraduate School 



(k ADDRESS (Cry. Stat*. and ZIP Code) 

Monterey, California 93943-5000 



7b ADDRESS (Ofy. Stste, and ZIP Code) 

Monterey, California 93943-5000 



3a NAME OF FUNDING/SPONSORING 
ORGANIZATION 



8b OFFICE SYMBOL 
(If applicable) 



9 PROCUREMENT INSTRUMENT IDENTIFICATION NUMBER 



8c ADORE SS (Ofy, Stafe. and ZIP Cod*) 



10 SOURCE OF FUNDING NUMBERS 



PROGRAM 
ELEMENT NO 



PROJECT 
NO 



TASK 
NO 



WORK JNIT 
ACCESSION NO 



ii title (include Security Claudication) 

SPUTTERING OF CHEMIS0RBED NITROGEN FROM THE (100) PLANES OF TUNGSTEN AND MOLYBDENUM: 

A COMPARISON OF COMPUTER SIMULATION AND EXPERIMENTAL RESULTS 



•: PERSONAL AUTHOR(S) 

Mattson. Philin J, 



' 3a TYPE OF REPORT 

Magr-pr'q Thpqi^ 



13b TIME COVERED 
FROM TO 



14 DATE OF REPORT (Year, Month Day) 

1986 December 



IS PAGE COuNT 



86 



'6 SUPPLEMENTARY NOTATION 



COSATi COOES 



F EiD 



GROUP 



SUB-GROUP 



18 SUBJECT TERMS (Continue on reverie if neceisary and identify by block number) 

Sputtering 
Computer Simulation 
Chemically Reacted Surfaces 



'9 ABSTRACT (Continue on reverie if neceuary and identify by 

The Naval Postgraduate School simulation 
nitrogen from the (100) faces of single 
placement was varied, and analyses were 
nitrogen. The cases where the adatom wa 
sputtered due to the collision cascade p 
were conducted to compare the results wi 
forts of earlier studies completed at th 
placement of nitrogen at 0.245 A from th 
similar to those found by Winters. The 
in that a substrate of greater mass resu 
agreed with Winters' findings, and confl 
The adatoms apparently reduce the moment 
inq the sputter yield ratio of the subst 



block number) 

model, QDYN86, was used to examine sputtering of 
crystals of molybdenum and tungsten. The nitrogen 
conducted on the sputtering cross sections of the 
s directly hit by the incident ion, or if it was 
rocess, were analyzed separately. The simulations 
th Winter's recent work, and to build upon the ef- 
e Naval Postgraduate School. It was found that 
e surface of molybdenum resulted in cross sections 
effect of the mass of the substrate was verified, 
Its in a higher sputtering cross section. This 
icted with earlier conclusions of past theses, 
urn available to create collision cascades, reduc- 
rate when the ions directly hit the adatoms. 



10 O S'R'BUTlON/ AVAILABILITY OF ABSTRACT 

£3 .:NCLASSlFiED^UNLiMlTED □ SAME AS RPT D DTiC USERS 



21 ABSTRACT SECURITY CLASSIFICATION 

UNCLASSIFIED 



22a NAME OF RESPONSIBLE 1ND1V1OUAL 

Don E. Harrison, Jr. 



22b TELEPHONE (Include Area Code) 

(408) 646-2877 



22c OFFICE SYMBOL 

Code 61Hx 



ODFORM 1473,84 mar 



83 APR edition may be used until e»hausted 
All other editions are obsolete 



SECURITY CLASSIFICATION! OF "his PACE 

UNCLASSIFIED 



Approved for public release; d i str i Put i on is unlimited 



Sputtering of Chemisorbed Nitrogen from the (100) Planes of 
Tungsten and Molybdenum: A Comparison of Computer Simulation and 

Experimental Results 



by 



Ph i I ip Jay Mattson 
Captain, United States Army 
B.S., Oregon State University, 1977 



submitted in partial fulfillment of the 
requirements for the degree of 



MASTER OF SCIENCE IN PHYSICS 

from the 

NAVAL POSTGRADUATE SCHOOL 
DECEMBER 1986 



ABSTRACT 

The Naval Postgraduate School simulation model, QDYN86, was used to 
examine sputtering of nitrogen from the (100) faces of single crystals 
of molybdenum and tungsten. The nitrogen placement was varied, and 
analyses were conducted on the sputtering cross sections of the 
nitrogen. The cases where the a da torn was directly hit by the incident 
ion, or if it was sputtered due to the collision cascade process, were 
analyzed separately. The simulations were conducted to compare the 
results with Winters' recent work, and to build upon the efforts of ear- 
lier studies completed at the Naval Postgraduate School. It was found 
that placement of nitrogen at 0.245 A from the surface of molybdenum 
resulted in cross sections similar to those found by Winters. The 
effect of the mass of the substrate was verified, in that a substrate of 
greater mass results in a higher sputtering cross section. This agreed 
with Winters' findings, and conflicted with earlier conclusions of past 
theses. The adatoms apparently reduce the momentum available to create 
collision cascades, reducing the sputter yield ratio of the substrate 
when the ions directly hit the adatoms. 






TABLE OF CONTENTS 



I . BACKGROUND 7 

A. HISTORICAL OVERVIEW 7 

B. PHYSICAL UNDERSTANDING 10 

C THEORETICAL AND EXPERIMENTAL DEVELOPMENTS 15 
(1960 TO PRESENT) 

D. COMPUTER SIMULATIONS 16 

E. APPLICATIONS 18 
I I . OBJECTIVES 21 

A. PREVIOUS EFFORTS 25 

B. WINTERS' SINGLE CRYSTAL SPUTTERING EXPERIMENTS 25 

C. THESIS OBJECTIVES 27 
III. COMPUTER SIMULATION AND MODEL DEVELOPMENT 28 

A. THE COMPUTER MODEL AND RELATED PROGRAMS 28 

1 . QDYN86 28 

2. Ancillary Programs 30 

B. SUBSTRATE AND ADATOM PROPERTIES 31 

C. POTENTIAL FUNCTIONS 33 

1 . Genera I 33 

2. Selection of Potential Function Parameters 35 

D. TARGET AND IMPACT AREA CONCERNS 44 



IV. RESULTS AND DISCUSSION 47 

A. GENERAL 47 

1. Sputtering Cross Sections 47 

2. Winters' Results 47 

3. Previous Theses at the Naval Postgraduate School 48 

4. Analyses Conducted in This Study 49 

B. MOLYBDENUM RESULTS 50 

1. Nitrogen Sputtering Cross Sections and Adatom 50 
P la cement 

2. Nitrogen Sputtering Cross Sections as a 55 
Function of Energy 

3. Sputtering of the Substrate 58 

C. TUNGSTEN RESULTS 58 

1. Nitrogen Sputtering Cross Sections and Adatom 58 
P la cement 

2. Nitrogen Sputtering as a Function of Energy 66 

3. Sputtering of the Substrate 66 

D. COMPARISON OF MOLYBDENUM AND TUNGSTEN RESULTS 68 

1. Comparison of Cross Sections 68 

2. Comparison of Substrate Sputtering Yields 70 

E. DETERMINATION OF POSSIBLE MASS EFFECTS 70 

F. COMPARISON WITH PREVIOUS SIMULATIONS 74 

V. CONCLUSIONS AND RECOMMENDATIONS 75 
APPENDIX DETERMINATION OF SPUTTERING CROSS SECTIONS 77 
LIST OF REFERENCES 80 
INITIAL DISTRIBUTION LIST 85 



ACKNOWLEDGEMENT 

I wish to thank those who preceded me, Steve Webb and Dirk Meyer- 
hoff, and to acknowledge the fact that I drew quite heavily upon their 
efforts. I also wish to thank Professor Harrison for his patience, and 
the concern he showed throughout this exercise. Special thanks to my 
wife, Lyn, and my sons Karl and Curtis who put up with a lot during the 
final weeks of this project. Finally, special heartfelt thanks to Dana 
Majors, who's contribution to the effort could not go unrecogn i 2ed. 



I. BACKGROUND 

A. HISTORICAL OVERVIEW 

When a surface is bombarded by a beam of energetic particles under 
proper conditions, surface damage effects can be observed. This damage 
can manifest itself a number of different ways; as surface damage in the 
form of pits, blisters, and cones, and in the ejection of atoms from the 
target. This ejection of atoms during bombardment is called sputtering . 

Sputtering was first discovered in 1853 by Grove [Ref. 1] when he 
observed the disintegration of the cathodes in glow-discharge tubes. He 
noted that the cathode material was deposited on the glass walls of the 
tubes. He called this process cathode sputtering . This erosion of the 
cathode was also noted by Plucker [Refs. 2-4]. Gassiot [Ref. 5] and 
Faraday also reported similar observations. 

Fifty years passed before any explanation for these phenomena was 
proposed. In 1902 Goldstein [Ref. 6] presented evidence that the sput- 
tering effects reported by Grove and Faraday were caused by positive 
atoms of the discharge impacting on the metal cathode of the tubes. 
Seven years later, in 1909, Stark proposed two models in an attempt to 
explain sputtering. The "hot spot model" considered the sputtered atoms 
to be the result of evaporation of target material from a small surface 
region due to localized heating by the ion beam. The "collision model" 
proposed that sputtering events were the result of a series of binary 
collisions initiated by a single ion [Ref. 7]. These models served to 
explain the experimental results at the time. 

7 



Later, in 1921, Thompson [Ref. 8] contributed to the field by 
suggesting that the atomic ejection was caused by the release of radia- 
tion as the ion struck the target. In the following years, a consider- 
able amount of theoretical and experimental work was done. Bush and 
Smith [Ref. 9] in 1923 attempted to describe the ejection of atoms as 
the results of the expansion of gas adsorbed by the target material. The 
following year, Kmgdon and Langmuir [Refs. 10, 11] conducted an exper- 
iment that suggested a momentum transfer ejection mechanism for sputter- 
ing. They bombarded thonated tungsten with ions in a glow discharge 
tube. This was a special case of sputtering, where the thin surface film 
of thorium sputtered rather than the tungsten substrate. 

In 1926, von Hippie and Blechschmidt [Refs. 12-15] proposed a 
theory that described sputtering as an evaporation of the surface atoms. 
Through spectroscopic techniques, von Hippie found that some of the 
sputtered atoms were in an excited state. He made further refinements 
on Stark's hot spot model, and made the first attempt to formulate a 
sputtering theory on the basis of local heating. 

Lamar and Compton [Ref. 16], in 1934, published "A Special Theory 
of Cathode Sputtering," which led to the "thermal spike" concept. They 
suggested that binary collision processes were dominant in light-ion 
sputtering, where local evaporation predominates for heavy-ion sputter- 
ing. The thermal spike concept was based on momentum transfer between 
the bombarding ion and the lattice atoms. The theory suggested that a 
long lived high temperature volume persisted in the target after the 
collision cascade was completed. 



8 



One of the major problems facing experimentalists was the lack of 
ability to reproduce results. Penning and Moubis in 1940 published the 
results of their studies of the effect of pressure on sputtering yield 
[Ref. 17], (the sputtering yield is defined as the ratio of "sputtered" 
target atoms per incident ion). They found that the sputtering yield 
was reduced with an increase in pressure. Collisions between the 
ejected atoms and the surrounding gases at the higher pressures resulted 
m the bacKscattenng of ejected atoms back onto the target surface, 
reducing the sputtering yield. When the background pressure was kept 
below 10~5 Torr, they were able to obtain reproducible results for ion 
energies in excess of 500 eV. 

Keywell made an attempt to formulate Stark's collision model in 
terms of a neutron transport model originally developed for nuclear work 
[Ref. 18]. Wehner [Refs. 19 - 21], one of the major contributors to the 
modern understanding of sputtering, began publishing his findings in 
1954. He discounted the evaporation model, and presented strong evi- 
dence for a momentum transfer process. He demonstrated the effects of 
crystal structure on the yield, and it became apparent that local heat- 
ing alone could not account for the effects of sputtering. His observa- 
tions revived interest in collision theory. One of his major contribu- 
tions was the discovery of the anisotropic ejection patterns from mono- 
crystalline targets, the now well known "Wehner spot pattern." 

In 1956 Harrison [Ref. 22] applied statistical methods to sputter- 
ing. He developed a theory involving the interaction of two distribu- 
tion functions; one for the crystal lattice and one for the ion beam. 



He utilized transport theory unci introduced the idea of idea! collision 
cross sections to explain sputtering from amorphous materials. 

Sputtering, to this point, had been regarded as an annoying mani- 
festation, one that eroded cathodes and filaments, contaminated plasmas, 
and was a general nuisance. There was great difficulty in obtaining 
reproducible experimental results. The theories proposed had not com- 
pletely described the phenomena, and indeed have not yet today. Before 
continuing with the discussion of the current trends m sputtering 
theory and experimentation, a brief discussion of the currently 
"accepted" concepts of sputtering is in order. 

B. PHYSICAL UNDERSTANDING 

Sputtering is the ejection of atoms from a target when bombarded by 
energetic ion projectiles. The incoming ion collides with atoms in the 
bulk of the target material, transferring part of its kinetic energy 
and momentum to the target atoms. If the energy transferred to an atom 
is greater that the binding energy at the lattice site, then a pr imar y 
r eco i 1 atom is created. This recoil atom may then collide with other 
atoms in the lattice, creating a col 1 i s ion cascade , transferring energy 
throughout the material. A surface atom may be ejected (sputtered) if 
the normal component of the energy transferred to it is greater than the 
surface binding energy of the lattice [Ref. 23], Figure 1 illustrates 
three concepts of the sputtering process. Figure 1 (a) shows the inter- 
action of the incident ion beam with the surface layer of the target. 
Figure 1 (b) illustrates the concept of the "thermal spike," where a 
high local temperature results in the evaporation of surface atoms. 



10 



Finally, F.gure 1 (c) illustrates the collision cascade process, which 
is the current conceptual model for sputtering [Ref. 24). 



Ion Beam 



Surface 




(a) 



(b) 



(c) 



(a) Sputtering of surface atoms as a result of the inter- 
action of the incident ion beam and the target sur- 
face. 

(b) Surface atoms evaporated as a result of a thermal 
spike. 

(c) Sputtering from a collision cascade, where energy is 
directed toward the surface from multiple binary 
collisions. [Ref. 24] 



Figure 1. Models of the Sputtering Pro 



cess. 



11 



The mechanisms for the sputtering of multicomponent materials 
(targets consisting of more than one atomic component) are more compli- 
cated. One of the simpler cases is the situation where a thin layer of 
a third component is placed on the surface of the thick target material. 
This is similar to the case examined by Kingdon and Langmuir in Refer- 
ence 11. In this case, the added atoms on the surface are called adatoms 
and the bulk of the material is called the substrate [Ref. 25]. Examples 
of this situation were covered by Garrison, Wmograd, and Harrison 
[Ref. 26] for oxygen on copper, and by Winters and Sigmund for nitrogen 
on tungsten [Ref. 27]. 

Winters proposed three mechanisms for the sputtering of the surface 
atoms, illustrated below in Figure 2. Layer (3) is the thin layer of 
adatoms on the surface of the thick target (2). An ion (1) bombards the 
target surface. In Figure 2(a), the ion hits the adatom, and the adatom 
is reflected off the substrate, either directly, or after it penetrates 
the target slightly. Figure 2(b) illustrates the case where the ion 
does not hit the adatom directly, but penetrates the target, and is in 
turn reflected itself. The reflected ion then hits the adatom, which is 
then sputtered off. Figure 2(c) shows the case where the ion causes an 
outward collision cascade, much like that illustrated in Figure 1(c). 
The outward flux of sputtered substrate in turn sputters the adatom. 

The basic processes associated with sputtering are similar to those 
causing radiation damage in the bulk of a solid. Sputtering, however, 
usually involves atoms in the distorted surface layers ( selvage ) of the 
target material. The term, "selvage," is derived from the term "sel- 
vedge," which is a narrow band woven such that the edge will not 

12 



unravel. This analogy was extended to the distorted layers of the sub- 
strate material at its surface [Ref. 28]. 



(a) 



(b) 




(1 


) 




(3)V^ 




(2) 



(c) 



V / 




>rf(3) 




\ ( i^ 





Figure 2. Three Proposed Mechanisms for the Sputtering of a Thin 
Layer on a Thick Target [Ref. 27]. 



Sputtering events take place in conditions far from thermal equili- 
brium, and are not evaporation of the material. Several factors can 
influence the sputtering yield, including incident ion energy, angle of 
incidence, ion type, target material, target crystal orientation, the 
presence of adsorbed molecules on the surface of the target, ambient 



13 



pressure, and numerous other factors. In order to obtain reproducible 
experiments, the following conditions must be met [Ref. 29]: 



1. The target surface must be clean, that is free of contaminants in 
the form of adsorbed gases, lubricants from the vacuum pump, 
cleaning solvents, etc. 

2. The gas pressure must be such that the mean free path of ions and 
sputtered atoms is large. 

3. The ion current density must be high and the background pressure 
low so that formation of surface layers is prevented during the 
experiment. 

4. The ions must strike the target at a known angle. 

5. The energy spread of the incident beam must be small. 

6. The ionizing conditions in the ion source should be such as to 
minimize the production of multiply charged species; the ions 
must be uniformly charged and mass separated. 

7. The lattice orientation of monocrystallme targets must be known. 



The field of sputtering is rich with terms used to describe the 
various observed and theoretical aspects of sputtering. Specific terms 
will be introduced as necessary to describe certain events, but two 
terms do merit mention here. Transmission sputtering is the ejection of 
atoms from the rear of a thin target. This occurs when sufficient 
energy is transported throughout the target to allow atoms to overcome 
the potential energy binding it to the target, allowing it to escape. 
Back-sputtering is the more familiar form, where atoms are ejected from 
the surface of the target material. Two other terms widely used, espe- 
cially when discussing the sputtering of multicomponent materials are 
physical sputtering and chemical sputtering. Physical sputterin g 
involves a transfer of kinetic energy from the incident particle to the 
atoms in the target, and the subsequent ejection of the atoms. Back 

14 



sputtering and transmission sputtering are manifestations of physical 

sputtering. Chemical sputtering is the result of a chemical reaction 

induced by the bombarding particle, which produces an unstable chemical 
compound on the surface of the target [Ref. 30]. 

C. THEORETICAL AND EXPERIMENTAL DEVELOPMENTS (1960 TO PRESENT) 

In 1962, Wehner and others [Ref. 31] published data on the sput- 
tering yields of metals and semiconductors in the 100-600 eV range. 
Later in 1966, Wehner [Ref. 32], in a report for Litton Systems pub- 
lished detailed results covering years of low energy sputtering 
research. 

Silsbee [Ref. 33], in an effort to account for the angular distri- 
bution of ejecta, the "Wehner spot patterns," proposed a focused colli- 
sion model, which allowed the transport of momentum in crystals along 
preferred directions. Experimental results had indicated that the 
sputtering yield for single crystal targets were dependent upon the 
crystallographic orientation of the target and the incident ion beam. 
This seemed reasonable, considering the "holes" observed in crystal 
models of when viewed from different planes. Focusing can be seen to 
contribute to the development of collision cascades within the target, 
it could not explain the Wehner spot patterns. 

Sigmund and Lehmann [Ref. 34] proposed an alternative model, based 
on Boltzman transport equations, requiring the target surface to have an 
ordered surface. In this model, an atom in the collision cascade 
would sputter if its kinetic energy component normal to the surface was 



15 



sufficient to overcome the surface potential barrier. Thompson [Ref. 
35] proposed another model to account for the spot patterns, in which 
the surface attraction for the escaping atom causes a refraction of its 
velocity vector away from the normal, resulting in a distortion of the 
angular distribution of ejecta. The theories of Sigmund and Thompson 
predict the sputtering yields of amorphous or polycrystallme targets 
fairly well, but do not accurately reflect the experimental results from 
the ordered surfaces of single crystal targets. 

Numerous theories have been proposed, but none succeeds in fully 
describe all aspects of sputtering phenomena. Harrison [Ref. 36] in a 
recent review, stated that there were currently no less than seven types 
of sputtering theories in the literature. The statistical theories of 
Thompson [Ref. 35] and Sigmund [Ref. 34] predict the yield relatively 
well, and provide information on the ejected atom energy distribution 
function, and ejected atom angular distribution for polycrystallme 
targets. Kelly [Ref. 37] classifies sputtering processes according to 
time scales. 

During this time, computer simulations were used to study sputter- 
ing. The models and codes now in use have been evolving for the past 30 
years. The next section will present some basic philosophy of computer 
simulations, with applications to sputtering research. 

D. COMPUTER SIMULATIONS 

In 1960 a new tool was added to the scientific arsenal, aimed at 
examining sputtering. Gibson, Goland, Milgram and Vineyard [Ref. 36] 

published a report of the first computer simulation examining radiation 

16 



damage m a material. They simulated metallic copper and studied radia- 
tion damage events at low and moderate energies, up to 400 eV. In the 
model, one atom was given an arbitrary kinetic energy and direction of 
motion, simulating its having been struck by an energetic particle. 
This was one of the first published accounts of a computer simulation to 
study such events. 

The high operating speed of computers makes them the natural choice 
for processing the numerous calculations required in a numerical analy- 
sis. The use of a computer simulation frees one from the constraints of 
general theories, and allows one to examine the basic physics of the 
system, to see how theory and experiment interact. In the words of 
Harrison [Ref. 36, p. 4]: 

A simulation is not a theory: it is a mathematical tool which is used 
to test the fundamentals of a theory. The computer can model a 
system with a minimum set of physical assumptions. This inherent 
simplicity helps to elucidate complex problems like sputtering: 
Simulations develop ideas which can then be exploited by both 
experimentalists and theorists . 

Computer simulations fall into two general categories: time-step 
models and event-store models. An event-store program moves from event 
to event, skipping the intervening time. It maintains a list of poten- 
tial future events (hence the name "event-store"), and checks and 
updates the lists to determine which event happens next. This model 
works well when the model is well understood, and the events are well 
separated in time. A time-step program carries the model forward for a 
short period of time, computes everything that happens to the system m 
that, time interval, updates, and then continues on. This program is 
most useful when several things happen simultaneously. The time-step 



17 



program tends to be shorter than the event store program, but it also 
runs slower. 

The event-store model used in sputtering research is based on the 
binary collision approximation (BC) [Ref. 39]. The assumption made is 
that each particle only interacts with one other particle at a time, and 
this other particle is usually assumed to be stationary. These models 
are inherently linear calculations [Ref. 40]. 

The time-step model used in this investigation is based on simulta- 
neous multiple interactions (Ml). Newtons laws, usually expressed in 
Hamiltoman form, are numerically solved for many particles. The Ml 
model used in this thesis, QDYN86, is the latest revision of the Ml 
program developed and used at the Naval Postgraduate School, and Penn- 
sylvania State University. QDYN86 is a full-lattice simulation, which 
models the dissipation of an incident ion's momentum in a single crystal 
target, using classical mechanics. This program can generate different 
surfaces of several crystal structures. Adatoms can be placed on the 
surface of the crystal to simulate reacted surfaces. Specific details of 
the simulation will be presented in a later section of this thesis. 

As can be seen, a considerable effort has been expended to study 
the mechanisms of sputtering. There is a practical reason for the 
interest, beyond the lure of pure research. 

E. APPLICATIONS 

Sputtering was long regarded as an undesired and little-understood 
effect [Refs. 41, 42]. It destroyed cathodes and grids in gas discharge 



18 



tubes, and contaminated plasmas and the surrounding walls [Ref. 43]. 
There was great concern about damage to spacecraft and satellites from 
sputtering. Efforts to understand and reduce these effects provided 
major impetus to the study of sputtering. 

However; sputtering, in the form of the controlled removal of 
surface atoms from a target, is becoming especially important in its own 
right. The ability to precisely control an ion beam, and to remove 
atoms from surfaces is a very important tool in research, and in the 
manufacture of miniature components. 

Sputter ion sources can be used for cleaning of surfaces, by remov- 
ing adsorbed surface molecules to a degree that is impossible to achieve 
chemically or mechanically. The atoms sputtered from a surface can be 
analyzed in a mass spectrometer, giving information about the surface 
composition. One of the more important commercial applications of 
sputtering is the deposition of thin films on a large variety of sub- 
strates, especially useful in the manufacture of microelectronics. 
Sputtering is used in micromachimng and depth profiling of thin films 
[Ref. 23]. 

There are applications of erosionally modified surfaces in the 
fabrication of optical, magnetic and surface acoustical technologies. 
Areas include grating fabrication, magnetic bubble technology, ion 
polishing, and reactive ion-beam etching [Ref. 44]. Research is being 
conducted to examine possible biomedical applications of sputtering. 
These include surface modification of biomedical materials, using sput- 
tering for pathological discrimination, and with applications to 
implants and prostheses [Ref. 45]. 

19 



One of the first commercial applications of sputtering was the use 
of sputtering for the deposition of solid film lubricants [Ref. 46]. 
Sputtering had the advantage that it allowed for the deposition of a 
variety of materials, on a large variety of substrates. The sputtered 
films are very dense, and strongly adherent. These properties are 
particularly useful for corrosion resistance and lubrication. 

The list of possible applications could continue. The ability to 
understand and control the surface characteristics of a material to the 
atomic level is a very powerful tool. The majority of the work, espe- 
cially m regards to the applications, is the result of considerable 
experimental effort. If one could better understand the mechanisms 
underlying the observed manifestations, then experiments could be better 
designed, and applications could be more quickly realized. it is toward 
this end that this research is conducted. 



20 



II. OBJECTIVES 

A. PREVIOUS EFFORTS 

The interaction of nitrogen with the (100) plane of tungsten has 
been widely examined. It was considered to be a useful system on which 
to base the development of adsorption kinetic and dynamic models 
[Ref. 47]. Winters [Ref. 48], in 1982, published a paper comparing the 
sputtering of chemisorbed nitrogen from polycrystalline targets of 
molybdenum and tungsten. His experiments showed that the nitrogen 
sputtering yield tended to increase as the atomic weight of the target 
substrate increased. 

This paper of Winters provided the basis for two Masters theses at 
the Naval Postgraduate School. In 1983, Meyerhoff [Ref. 49] used an 
earlier version of Harrison's computer simulation sputtering model to 
study sputtering from nitrogen reacted (100) surfaces of tungsten and 
molybdenum targets. His goal was to compare the simulation results 
with the experimental results recently published by Winters in Refer- 
ence 48. Meyerhoff concentrated on the reported mass effects. He 
concluded that while there may be a mass relation, the distance of the 
adatom from the substrate had a much more profound effect on the sput- 
tering yield. 

In 1986, Webb [Ref. 50] sought to further examine the adatom pla- 
cement problem. Webb again used (100) surfaces of monocrystallme 
tungsten and molybdenum reacted with nitrogen and bombarded with argon 

21 



ions. He narrowed the range of the placement of the adatoms, and his 
results indicated that the adatoms were positioned above the (100) sur- 
face of the target plane. 

In all these cases, the nitrogen adatoms were assumed to be in the 
four-fold position, as shown below in Figure 3. Meyerhoff assumed the 
nitrogen atom was slightly above the surface of the target plane, in a 
position of equal distance from its five nearest neighbors. Webb exa- 
mined the nitrogen sputtering yield with the nitrogen placed at three 
different heights for molybdenum and two heights for tungsten. 



Adatom in 4-fold 
— Position 




(100) surface// 
</ 



Centered i 
Atom s 



Figure 3. Nitrogen Adatom Placed in Four-fold Position in BCC 
Unit Cell. 



The basis for the four-fold placement assumption is well estab- 
lished in the literature. An earlier work by Clavenna and Schmidt 
examined the interaction of N 2 with W(100) [Ref. 51]. They studied the 
binding states and condensation and desorption kinetics of nitrogen on 
W(100) using flash desorption spectrometry. They felt that the 



22 



nitrogen was in tne four-fold position, but were uncertain exactly 
where the nitrogen was placed vertically in relation to the target sur- 
face. Adams and Germer [Refs. 52, 53) also examined the adsorption of 
nitrogen on tungsten. They used a combination of Low Energy Electron 
Diffraction (LEED), f lash-desorption mass-spectrometry and contact 
potential measurements in their experiments. They refined their exper- 
iment by using the 1 ^N-isotope of nitrogen, in order to distinguish 
between nitrogen and CO present in the background gas. They felt that 
since the nitrogen atom was half the size of the tungsten atom, that it 
would sit m the well formed by the surface atoms of the substrate. 

Griffiths, Kendon, King and Pendry [Ref. 54] examined target atom 
displacement due to the introduction of the nitrogen adatoms. They 
performed as series of LEED experiments with various nitrogen coverages 
and beam energies. They found that a fractional coverage of approxi- 
mately 0.4 (defined as the ratio of the adatoms to the number of sub- 
strate atoms for a specific face of the target) was relatively stable. 
At higher coverages, of about 0.5, the nitrogen would be absorbed into 
the bulk at temperatures of about 1000 K. The 0.4 coverages were 
stable at this temperature. The model suggested by Griffiths, et. al. 
was a contracted domain structure , where the nitrogen occupies a four- 
fold hollow site, but that the four surface tungsten atoms are uni- 
formly displaced towards the adatom, This leads to a series of "is- 
lands" consisting of 16 nitrogen atoms and the corresponding surface 
tungsten atoms, as shown in Figure 4. This type of formation was phys- 
ically modeled by conducting a series of laser diffraction experiments. 
Laser diffraction gratings were constructed, with an average island 

23 



size of 4x4, and when illuminated by laser light produced diffraction 
patterns similar to those of the LEED beams on the actual target. 
Again, while this provided insight into the relative placement of the 
nitrogen on the surface, it did not provide any information on how the 
nitrogen was related vertically to the crystal surface. 




rtw^w 










Illustrates contracted-domain structure of the 0.4 
monolayer of Nitrogen on W(100) or Mo(100) surface. 
Large hatched circles illustrate the top layer of the 
substrate atoms, the small filled circles, the 
nitrogen atoms. 

a) Plan view: Shows Domain and Boundary of Structure. 

b) Cross Section through line AA. [Ref. 54, p. 1586]. 
Figure 4. Suggested "Island" Formation of Nitrogen on Tungsten. 



24 



Meyerhoff assumed the equilibrium position of the nitrogen in his 
research. Webb examined equilibrium points both above and below the 
substrate surface plane. While there was a great body of information 
available on the adsorption of nitrogen on single crystal surfaces, 
there was little data on the sputtering of nitrogen from single crystal 
surfaces. This was rectified by a recent report by Winters [Ref. 55]. 

B. WINTERS' SINGLE CRYSTAL SPUTTERING EXPERIMENTS 

This paper [Ref. 55] was a continuation of Winters' previous 
studies in the sputtering of chemisorbed nitrogen on tungsten. In his 
1974 paper he proposed two mechanisms for the sputtering of nitrogen 
from the tungsten surface [Ref. 27]. He suggested that for low energy 
incident ions, the primary mechanism for the sputtering of the nitrogen 
was direct knock-on collisions with the impinging or reflected bombard- 
ing ions. The proposed mechanism for sputtering at higher bombarding 
energies included nitrogen atoms knocked away by sputtered substrate 
atoms. Winters' 1982 paper [Ref. 48] examined the sputtering of 
nitrogen from molybdenum and tungsten polycrystallme targets. 

This latest report by Winters and Taglauer described experiments 
investigating the sputtering of chemisorbed nitrogen from W(100), 
W(111), W(110) and Mo(100) single crystal surfaces. The targets were 
bombarded with helium, argon and xenon ions in the energy range of 300 
eV to 5000 eV. The relation between the sorbate mass, substrate mass 
and sputter yield were examined. Conclusions were drawn about thermal 
spikes, recoil implantation, and cascade mixing. Since these exper- 
iments were conducted on single crystal surfaces they provided some 

25 



basic information on the physics of the sputtering of multicomponent 
systems. Additionally, since the experimental data wab derived from 
single crystals, they lend themselves more directly to computer model- 
ing. A brief description of the experiment indicates how the system 
can be modeled in the computer simulation. 

A Faraday cup and the four single crystal targets, W(100), W(111), 
W(110), and Mo(100) were mounted on a rotating manipulator. The tar- 
gets were cleaned by heating to about 2500K for tungsten and 2200K for 
molybdenum by RF induction heating [Ref. 56], The sample was exposed 
to 15 N 2 until about 1/2 monolayer was adsorbed. This is in agreement 
with the findings of stable nitrogen adsorption on tungsten as men- 
tioned m references 52 thru 54. The use of the 15 N isotope allowed 
one to isolate the adsorbed species involved in the study. 

After the adsorption, an Auger spectrum was taken to ensure the 
absence of impurities from the sample surface. The ion beam was 
scanned by the Faraday cup in order to ensure beam uniformity, and to 
measure the beam current density. The ion beam was rastered across a 
0.48 cm aperture, impinging at normal incidence upon the sample cen- 
tered behind the aperture. A nitrogen Auger peak-to-peak intensity was 
determined as a function of the ion dose. Finally the sample, which 
was partially cleaned, was exposed to the ambient for approximately 600 
sec after each run to monitor the readsorption of nitrogen. If read- 
sorption was observed, the run was discarded. 

The care taken in these experiments is quite apparent. The 
parameters of this experiment lend themselves quite well to computer 
simulation. Winters' also presented a comparison of the experimental 

26 



results with theoretical calculations based on the Sigmund-Wmters 
theory for the sputtering of chemisorbed gas [Ref. 27]. One recommend- 
ation in the paper was : 

In particular, it would be useful to compare an extensive set of 
molecular dynamics calculations with both the experimental data and 
also the calculations presented in this paper. [Ref. 55, p. 26] 

C. THESIS OBJECTIVES 

The primary objective of this thesis is to examine sputtering from 
the (100) surfaces of single crystals of molybdenum and tungsten. 
Nitrogen will be placed in different equilibrium positions in relation 
to the substrate surface, and the system will be bombarded with argon 
ions of various energies at normal incidence. The effect of the mass 
of the substrate in relation to the sputtering yield will be examined. 



27 



III. COMPUTER SIMULATION AND MODEL DEVELOPMENT 

A. THE COMPUTER MODEL AND RELATED PROGRAMS 
1 . QDYN86 

The computer simulation used in this study is called QDYN86, 

which stands for the 1986 revision of the QDYN (Quick DYNamics) program 

used by Harrison at the Naval Postgraduate School. This program uses 

multiple-interact ion (Ml) logic in a t ime-step approach. The time-step 

logic is appropriate when many events occur simultaneously. The model 

is based on classical mechanics, using Newton's laws expressed in HamM- 

toman form to simplify the calculations. The initial inputs to the 

program include the target crystalline structure, atomic masses and 

potential functions of the substrate, incident ion, and any adatoms; 

adatom locations, and bombarding ion angle of incidence, energy and 

impact point. The program develops the subsequent collision cascade, 

tracking the positions and velocities of the target atoms through time. 

The computations terminate when there is insufficient energy for any 

further ejections to occur. 

The positions and velocities of the moving atoms are tracked 
through each time-step. In order to strike a compromise between exces- 
sive computer run time, and yet to maintain reasonable energy conserva- 
tion, the time-step increment is variable. The time-step is determined 
by a specified distance divided by the highest atomic velocity in the 

28 



target. The distance chosen is 0.1 lattice units (LU) , where the lat- 
tice unit is defined as one-half the lattice parameter, a (for a cubic 
structure). The time-step increment is further modified by other fac- 
tors which take into consideration the previous velocity, smoothing the 
trans it ion in time. 

The velocities and positions of every atom in the target are 
not calculated for each atom at every time-step. Again, this would 
consume excessive amounts of computer time. As a result, the forces are 
not computed on a specific target atom until the a torn is struck by a 
moving atom. Atoms which rise above the target surface are put m a 
tentative list of ejected atoms. After further atom ejection from the 
target is unlikely, the atoms in this list are tested to see if they 
were travelling with sufficient velocity to overcome the attractive 
forces of the target atoms. Atoms which fail this test are added to the 
target, and are not counted as having been sputtered. Specific actions 
are taken for atoms that leave the bottom of the target, and through the 
sides. The program ma inta ins , and periodically updates a listing of an 
atoms nearest neighbors. After the completion of a collision cascade, 
the program reinitializes, and prepares for another trajectory. A num- 
ber of trajectories are run at different impact points in order to 
obtain better statistical results. 

Subsequent sections in this chapter will provide further 
details on the choice of potential functions, substrate characteristics, 
impact points, and other factors as they pertain to this study. A num- 
ber of other articles provide more detailed information on the QDYN 
simulation, and other molecular dynamics simulations. Reference 36 

29 



provides a good overview of sputtering models. Harrison and Jakas com- 
pared the Ml and binary collision models [Ref. 57], and with Webb, 
examined a hybrid code between the two models [Ref. 58]. The most 
recent, and most detailed discussion of the QDYN program by Harrison and 
Jakas identified uses of the program for studies involving insulators 
and semiconductors [Ref. 59]. 
2 . Anc i 1 lar y Programs 

A number of other programs are used to support the mam simu- 
lation. These programs are identified and are briefly described below. 

a. QDYNLIBB (QD86LIBB) 

This program forms a library from which the main program 
derives routines for calculating the forces and potentials. The poten- 
tial and forces tables are dimensioned to 10000, and support four poten- 
tial functions. Ada torn routines are available for placing adatoms on 
the surface of the target, or creating steps or vacancies. 

b. TARGLIBB 

This program forms a library of routines used for generat- 
ing (100), (110), and (111) faces of BCC, FCC and Diamond lattice tar- 
gets. A rotated FCC (001) lattice can also be generated. 

c. ANMOL 

This program analyzes the basic data set generated by the 
mam program. The output of the mam program consists mainly of a list- 
ing of the positions and velocities of the atoms. ANMOL uses this data 
file to calculate which atoms are sputtered, from which layer in the 
target lattice they originated, and rotational and vibrational analysis 



30 



of the ejected atoms. Determination of mu 1 timers, ejection t ime cj i s 
tri but ions and ejection energies are also calculated. 

d. ANPLOT 

This is a program used to interface with the DISSPLA 
graphics package. ANPLOT is used to generate graphical analyses of the 
data file generated by the mam program. 

e. POTTER 

This program is used to generate potential functions. An 
abbreviated input deck is read by POTTER, and two data files are gener- 
ated. These files, FORCE DATA and ENERGY DATA, are used by the next 
program to obtain a graphical representation of the functions. 

f. POTPLOT 

This is another graphics routine that interfaces with the 
DISSPLA graphics package. POTPLOT uses the data files generated by 
POTTER, and produces a liner plot of the potentials, and a semi-log plot 
of the forces. This allows one to quickly determine which modifications 
must be made to obtain the proper shaped curve for the desired potent ni 
and force function. The potential functions used in this thesis were 
generated using this program. 

B. SUBSTRATE AND ADATOM PROPERTIES 

The systems of n i trogen-on-mo 1 ybdenum and n i trogen-on-tungsten were 
chosen by Winters for his sputtering experiments for a very specific 
reason. Many of the physical parameters of the two substrate materials 
are very similar, and the behavior of nitrogen with each was thought to 
be almost identical also. This provided a means of physically reducing 



31 



the variables in the system, enabling him to isolate and concentrate on 
the mass difference of the two materials. A summary of the Key physical 
parameters for the two systems is show below in Table 1. 

TABLE 1. PHYSICAL DATA FOR MOLYBDENUM AND TUNGSTEN. 

MOLYBDENUM TUNGSTEN 



Atomic Number (Z) 42 74 

Atomic Weight (amu) 95,94 183.85 

Atomic Radius (A) 1.363 1.371 

Density (gm/cm 3 ) 10.22 19.3 

Crystal Type BCC BCC 

Lattice Constant (A) 3.147 3.165 

Lattice Unit (1/2 a , A) 1.5735 1.5825 

R e (A) 2.8 2.894 

Valence +4 +4 

ionic Radius (A) 0. 70 0.70 

Cohesive Energy (ev) 6.82 8.90 

Binding Energy of N (eV) 6.5 6.5 

Data for Table 1 derived from References 60-62. 



Table 1 illustrates the high degree of similarity between some of 
the physical properties of molybdenum and tungsten. Nitrogen is assumed 
to occupy the four-fold position on the surface of both materials. The 
major difference between the two materials is the significant difference 
in mass and Z. The pertinent physical properties of the adatom, 
nitrogen; and the incident ion, argon; are summarized below in Table 2. 

TABLE 2. PHYSICAL DATA FOR NITROGEN AND ARGON. 

Nitrogen Argon 

Atomic Number (Z) 7 18 

Atomic Weight (amu) 15 39.94 

Data for Table 2 derived from Reference 63. 

32 



C. POTENTIAL FUNCTIONS 
1 . Genera 1 

The dynamics of the atomic interactions are controlled by the 
choice of the potential functions used to describe the potential 
energies and forces felt by the atoms. There is no "right choice" for a 
potential function in these types of simulations, indeed, Harrison has 
called the development of useful potential functions for simulations a 
"black art". A number of different potential functions can and have 
been used in sputtering simulations. Harrison has found that the choice 
of potential function parameters do not overly affect the outcome of the 
simulation [Ref. 64]. The potential functions used in this investiga- 
tion are described briefly below. Detailed discussions of these, and 
other potential functions can be found in Torrens [Ref. 65]. A number 
of survey articles by Harrison describe more fully how the potential 
functions are applied to the simulations [Refs. 36, 57-59]. 

a. Born-Mayer 

The Born-Mayer potential function is a purely repulsive 
function. This function is used for intermediate atomic separations. 
The form of the function is 

V(r) = a exp(-br) . ( 1 ) 

b. Mo 1 i ere 

The Moliere potential function is another purely repulsive 
potential function, of a slightly different form. It is called a 
"Screened Coulomb Potential," and is an approximation of the Thomas- 
Fermi screening function. The general form of the potential is 

33 



V(r) = (Z 1 Z 2 e 2 / r)[0.35 exp (-0.3r/a) + 0.55 exp(-1.2r/a) + 

0. 1 exp (-6.0 r/a) ] . ( 2 ) 

The term "a" is called the "Firsov length" defined as follows: 

a r k 0.8853 a b /{Z s 1 / 2 + Z 2 1 /2)2/3 , ( 3 } 
The " a ^ " term is the Bohr radius, 

a b - h 2 / 4tt 2 me 2 - 0.5292 A. ( 4 ) 

The "a" parameter may be modified, since the Moliere potential is an 

approximation to the Thomas-Fermi function. When k = 1, the potential 
is pure Moliere; when K is set unequal to 1 then the function is known 
as a "modified Moliere" potential function, 
c. Morse 

The Morse potential is both attractive and repulsive, 

depending upon the separation distance, "r". It has the form: 



V(r) - D e exp [-2 a (r-r e )] - 2D e exp [- a(r-r e )]. ( 5 ) 



The term, D e is the well depth, and r e is the equilibrium separation of 
an atomic pair, and alpha is a scale factor which controls the shape of 

34 



the potential well. The function is attractive for r > r e , and repul- 
s i ve for r < r e , 

d. Composite Mor se-Mol iere 

The composite Morse-Mo I i ere potential function is used to 
more closely model the dynamics of the sputtering events. The repulsive 
wall of the Morse potential is joined with a cubic spline to the Moliere 
potential, to form a single potential function which can be used over 
long ranges. The two functions are joined by varying the alpha term of 
the Morse, and the "a" term of the Moliere to obtain an intersection 
which has a smooth, continuous slope. The repulsive wall governs the 
collision dynamics, and the attractive well of the Moliere controls the 
sputtering by determining whether an atom will escape the surface of the 
target. 

2. Selection of Potential Function Parameters 

a. Substrate-Substrate, Adatom-Adatom Function Parameters 

A composite Mor se-Mo 1 i ere function was used for the solid 
phase Mo-Mo and W-W potential functions. The specific parameters used 
m this thesis follow those developed by Webb, with one exception. The 
function used for the W-W forces had a discontinuity as shown in Figure 
5. This was smoothed by changing the R a value. The potential curves 
are shown in Figure 6. 

The nitrogen-nitrogen potential function is a pure Morse 
potential function. The tabulated data for the solid phase W-W, Mo-Mo, 
and N-N potential functions is listed below in Table 3. 



35 



3 
O 

o 

3 -• 



O 

o 

o 




o -• 



o-- 



o 

rv, 



O 

o 

O 

o 




o 
o 

d 



o 

o 
o 
o 

o. 
o 
o 
q 

d J. 



TT7 



MO/ 



0.25 0.50 0.75 1.00 

SEPARATION (LU) 



1.25 



1.50 



Figure 5. w-w and Mo-Mo Forces 



36 




i r 

0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 



SEPARATION (LU) 



Figure 6. W-W and Mo-Mo Potentials Generated By Webb. 



37 



TABLE 3. SOLID PHASE POTENTIAL PARAMETERS 





(eV) 


R e 

(A) 


a 
(A" 1 ) 


k 


Ra 
(LU) 


Rb 
(LU) 


R c 

(LU) 


Mo-Mo 

W-W 
N-N 


0.997 
1 .335 
7.373 


2.800 
2.894 
1 .098 


1 .519 
1 .200 
2.700 


0.0 
0.0 
0.0 


0.790 

0.80 

0.0 


0.830 
1 . 130 
0.0 


2.40 
2.40 
1 .71 



The a and k terms were used to match the slopes of the 
Morse and Moliere potential functions, as mentioned earlier. The value 
of D e was selected in order to obtain the proper cohesive energy for the 
substrate as shown in Table 1. The R e term is the nearest neighbor 
distance. The spline boundaries, R a and R b , are in turn a function of 
the a and k values selected, and should be no more than y t LU apart. This 
topic will be covered in more detail in the following section. The 
final value, R c , is a distance over which the force and potential 
calculations are made, a cut-off distance. This allows for interactions 
only between nearest and next-nearest neighbors. This value is set at 
1.71 LU for all interactions except the substrate-substrate interac- 
tions, where it is set. at 2.40 LU. 

b. MO-N, W-N Potential Function Parameters 

Six new potential functions were generated to examine the 
effect of adatom placement on the sputtering cross section. The FORTRAN 
routines POTTER and POTPLOT were used to help generate these functions. 
The first step was to select an elevation to place the adatom, and from 
that to determine R e , the nearest neighbor separation distance. R e is 
the distance measured vertically between the cubic center and the ada- 
tom, as shown in Figure 7. These values were then input to POTTER. 

38 



POTTER generates two data files used by the POTLOT routine 
to plot the potential and forces curves. The R a , R^, k, and alpha 
values are modified so that a smooth curve is plotted for both the 
potentials and forces. Once this rough determination is completed, then 
D e is modified and evaluated by running a single trajectory at one 
time-step under CMS in order to obtain the proper sublimation energies. 
D e is modified so that the sublimation energy of the nitrogen to the 
substrate is maintained at 6.5 eV. The binding energy of the adatom is 
one-half the sublimation energy, or 3.25 eV. 

After the sublimation energy has been properly set, then 
the revised values for D e are once again run through POTTER and POTPLOT 
in order to insure that the curves have the necessary smooth slopes. 
These graphical methods for rapidly evaluating potential function par- 
ameters <r\a\/e greatly eased the process for generating potential func- 
tions. The values for the Mo-N and W-N potential functions used in this 
thesis are listed below in Tables 4 and 5. 



Adatonru 



datom Placement Height 
Substrate 




Figure 7. Physical Dimensions Used in Potential Function 
Ca 1 cu 1 at ions . 



39 



TABLE 4. Mo-N POTENTIAL FUNCTION PARAMETERS 



Locat ion 
(A) 



(ev) 



(A) 



a 
(A) 



"a 
(LU) 



(LU) 



(LU) 



-0.05 
0. 146 
0.245 
0.290 
0.335 
0.38 



2 
2 
1 



970 

0428 

997 



1 .747 



672 
60 



1 .524 
1 .720 
1 .738 
1 .864 
1 .908 
1 .950 



2 

2 
2 
2 
2 

2 



60 

20 

143 

20 

20 

15 



0.80 

0.80 

0.80 

0.0 

0.0 

0.0 



0.540 
0.540 
0.500 
0.560 
0.520 
0.720 



0.560 
0.560 
0.520 
0.520 
0.560 
0.750 



71 

71 
71 
71 
71 
71 



** 



x* 



* • 



Note : Rows marked with ** indicate parameters developed by Webb, 
and used in both theses. Unmarked rows indicate new potential func- 
tion parameters. 



TABLE 5. W-N POTENTIAL FUNCTION PARAMETERS 



Locat ion 
(A) 



(eV) 



"e 

(A) 



a 
(A) 



-1 



"a 
(LU) 



(LU) 



(LU) 



-.05 
0. 1623 
0.2464 
0.325 
0.487 



3.292 
2.341 
2.013 
1 .774 
1 .712 



1.524 
1 .7448 
1 .8289 
1 .9075 
1.960 



2.85 
2.75 
2.60 
2.45 
2.32 



0.8 
0.0 
0.0 
0.0 
0.9 



0.65 
0.75 
0.80 
0.85 
0.83 



0.67 
0.80 
0.85 
0.90 
0.85 



1 .71 
1 .71 
1 .71 
1 .71 
1 .71 



* * 



x* 



Note : Rows marked with ** indicate parameters developed by Webb, 

func- 



and used in both theses, 
tion parameters. 



Unmarked rows indicate new potentia 



40 



c. Ion-Substrate, lon-Adatom Functions. The potential func- 
tion parameters for the incident ion must be considered. The target is 
bombarded by Argon ions, but for the purposes of these calculations, the 
charge on the ion is neglected. The ion is charged so that it can be 
accelerated to the desired energy to bombard the target. Since the 
incident ion is a noble gas, the event that it will react with the 
molybdenum, tungsten or nitrogen atoms other that through simple colli- 
sion process is unlikely. The potential functions are correspondingly 
modified Moliere. The potential function parameters are listed below in 
Table 6. 



TABLE 6. AR-N, AR-MO, AND AR-W POTENTIAL PARAMETERS 





De 
(eV) 


R e 

(A) 


a 

(A)" 1 


k 


Ra 

(LU) 


Rb 
(LU) 


R C 
(LU) 


Ar-N 

Ar-Mo 

Ar-W 


0.0 
0.0 
0.0 


0.0 
0.0 
0.0 


0.0 
0.0 
0.0 


0.0 
0.0 
0.0 


1 .71 
1 .71 
1 .71 


1 .71 
1 .71 
1 .71 


1 .71 
1 .71 
1 .71 



Figure 8 shows the Nitrogen-Nitrogen potential function, and 
Figure 9 shows the Ar-N, Ar-Mo, and Ar-W potential functions. 

d. Vacuum Phase Potentials 

Once atoms have sputtered from the surface of the target, 
there is a probability that further interactions can occur in the vacuum 
phase. Vacuum phase interactions were not emphasized in this study, but 
for the sake of completeness, the vacuum phase parameters derived in 
Webb's thesis are included in Table 7. There is little available data 

41 



N-N POTENTIAL 



o 

o ■ 

CJ 



o 

CO 
M 



© 

CJ 



© 
f~ -- 



-J " 



c 
c 



o 



© 

CJ 



0.0 



0.5 



10 1.5 

SEPARATION (LU) 



2.0 



:.5 



Figure 8. Nitrogen-Nitrogen Potential Function. 

42 



100 r 




2 

Separation, (A) 



Figure 9. Ar-Mo, Ar-N, and AR-W Interatomic Potential Functions 



43 



for the Mo-Mo, W-W, Mo-N, and W-N systems. The following substitutions 
were made, using similar systems to those examined in this study. The 
Ag-Br system was used fro Mo-Mo, and VO was used for Mo-N. Data on the 
N-N system was available and used. [Refs. 50, 65]. 

TABLE 7. VACUUM PHASE POTENTIAL FUNCTION PARAMETERS 

D e a R e Q Qx 

(eV) (A)" 1 (A) 



Mo-Mo 1.700 1.790 2.390 247.000 0.679 

N-N 5.760 3.359 1.098 2358.570 14.320 

Mo-N 6.100 1.965 1.524 1020.000 4.700 

W-W 3.100 1.450 2.470 308.399 0.960 

W-N 4.900 2.159 1.738 967.000 4.850 

RHC 2.900 1.560 2.390 1050.000 5.000 ** 

Note: The Rhodium vacuum data was used in some cases. 



D. TARGET and IMPACT AREA CONCERNS 

One characteristic of Ml simulations of this type, is that finite 
target sizes are used. Two concerns must be balanced in determining the 
optimum target size. First one must try to have a target of sufficient 
size such that the majority of the sputtering events are contained. If 
the target is too small, then information will be lost due to a f a i lure 
of conta i nment event. On the other hand, an excessively large target 
will require excessive computer time to complete each trajectory. In 
this thesis, two target sizes were used. At the 500 eV energy level, a 
19x6x19 target was used, and at higher energy levels a 23x8x23 target 
size was used. Webb used the larger target in his thesis. 

Nitrogen atoms were placed in the four-fold position on both target 
sizes. 40 atoms were placed on the smaller target, and 60 atoms were 



44 



placed on the larger. In both cases this resulted in a fractional cov- 
erage of approximately 0.4, which corresponds to the contracted domain 
structure of Reference 54. The nitrogen adatom coverage is shown in 
Figure 10, corresponding to the (001) face of the substrate material. 
Table 8 lists target area and nitrogen coverage parameters. 

TABLE 8. TARGET AREA AND ADATOM COVERAGE PARAMETERS 

Target Size 19 x 6 x 19 23 x 8 x 23 



Substrate Mo W Mo W 



No. Nitrogen 40 40 60 60 

Area (cm 2 ) 8.02x10~ 14 8.11x10" 14 1.20x10 -13 1.21x10~ 13 
Fract iona I 

Coverage 0.4 0.4 0.42 0,42 
Coverage 

(atoms/cm 2 ) 4.99x10 14 4.93x10 14 5.0x10 14 4.96x10 14 



Another factor which impacts on the run time of the simulation, and 
the statistical accuracy of the results is the number of impact points, 
or trajectories run. The impact points are designated by a pre-selected 
"pm-pomt", and then the trajectories impact in a representative area 
based on this point. Two types of pin-points were used. In one, the 
"hit" case, a nitrogen adatom is placed in the representative area, and 
can be hit directly by the incoming ion. The other, "non-hit" case, 
does not have a nitrogen atom in the representative area. The simula- 
tions were run in this study with 150 trajectories per pin-point, or 300 
trajectories per complete run. Previous studies used 300 trajectories 
per pm-point, for a total of 600 trajectories per case. 

The pin-points for the "hit" case were located at (8.000, 6.000) 
for the 19x6x19 target lattice, and (13.000, 15.000) for the 23x6x23 

45 



target. The non-hit cases were run with pin-points at (8.000, 8.000) 

and (11.000, 11.000) respectively. These pin-points are measured on the 

x- and z- axes. The representative areas and pm-points are also shown 
i n F igur e 10. 




Small circles represent nitrogen atoms. 
"H" indicates "Hit" representative area. 
"NH" indicates non-hit representative area 



Figure 10. Top View of Nitrogen Reacted Surrace, 



46 



IV. RESULTS AND DISCUSSION 

A, GENERAL 

i . S p u t X g r t n g c r o s s S e c 1 1 o n s 

The concept of the sputtering cross section is key to the 
analyses in this study. The derivation of the cross section is carried 
out in detail in Appendix A. The basic methodology used by Winters was 
adapted to the simulation, and the following relation was derived; 



(7 



N = " [lh (1 - Y/N ) } A . ( 6 ) 



Y is the sputter yield from the simulation, a^ is the Nitrogen -.putter- 
ing cross section, N is the initial number of adatoms placed on the 
target and A is the surface area of the target, 
£. Winters' Results 

Winters' experiment involved the bombardment of the (100) 
Plane of molybdenum and tungsten with argon, xenon and helium ions in 
energies from 300 ev to 5 kev. The cases studied in this thesis con- 
cerned argon ions of the 500 ev to 2 kev range on Mo (100) and W(100) , 
Winters' presented results from his experiments, and theoretical calcu- 
lations from the 8 igmund-W inters theory of mult ^component sputtering. 
The results that fall within the bounds of this study ar^ tabulated 
below in Table 9 [Ref. 55 p. 29]. 



47 



TABLE 9. WINTERS' THEORETICAL AND EXPERIMENT CROSS SECTIONS 
AS A FUNCTION OF ION ENERGY 



I on 

Energy 

(ev) 



Exper iment 
MO -16 cm 2 ) 

MO (100) W(100) 



ca i cu l* 


1 1 on 


( 1 o~ 1 6 


cm 2 ) 


MO (100) 


W(100) 


6.0 


6. 9 


6,9 


7,4 


P. .0 


8. 1 


6.9 


6-6 



500 
1000 
2000 
3000 



5. 6 


6. 6 


6,5 


9.6 


6. 8 


9.0 


6. 6 


6.6 



3. Previous Theses at the Naval Postgraduate School 

a. Meyer ho ff 

Meyer hoff's thesis m 1983 examined clean and nitrogen 
reacted molybdenum and tungsten (001) surfaces bombarded with argon ions 
of energies ranging from 500 ev to 3 Kev, Nitrogen was placed in the 
four-fdd position, and was examined at two elevations above the (OOi) 
surfaces, The effect of the mass of the substrate was examined by 
replacing the Mo-mass with the tungsten mass. Meyerhoff concluded that 
the sputtering cross section of the nitrogen was more dependant upon the 
distance of the adatoms above the surface then upon the mass of the sub- 
strate. [Ret, 49] 

b. webb 

Webb, in early 1966, continued the study of nitrogen on 
molybdenum and tungsten. He refined the potential functions used by 
Meyerhoff, and performed basic sputtering research using the QDYN simu- 
lation. He studied the sputtering cross section of nitrogen for place- 
ment at three different heights in relation to the (001) surfaces of 



48 



mo I /bdenum Jrld f.ungSten . He Concluded thdt the n i ' I'OQ^ 11 w,is rt k j 1 1 I - ' :■ i / 

located above the surface, as opposed to being in the well formed by the 
surface atoms, 

4 . Analyses Conducted in This Study 

The scope of the analyses conducted in this study differed 
somewhat from previous theses. The previous studies did not differen- 
tiate between the effects of the incident ion initially hitting the 
adsorbed nitrogen, or the substrate, The simulations were run with 
impact points covering both cases, but the results were not looked at 
separately. In this study, the two cases were analyzed separately. 
This was done in order to gain insight into possible sputtering mecha- 
n i sms as out l i ned ear l i er . 

The nitrogen sputtering cross sections were examined at 500 ev 
(low energy sputtering) and 2 keV (high energy sputtering). The hit and 
no-hit cases were examined in order to determine the sputtering mecha- 
nism of the mtrogen. The yield of the substrate was examined, and the 
ratio of the bare to the reacted substrate yields was used as a measure 
to determine the relative effects of the adsorbed mtrogen. The purpose 
of this analysis was to determine if the nitrogen enhanced the sputter- 
ing yield of the nitrogen, particularly at the higher ion energies. 

The effect of the mass of the substrate was examined, in order 
to attempt to resolve the difference between the findings of Winters and 
Webb and Meyerhoff. The technique used by Meyerhoff was again used in 
this study, except that the sputtering cross section of nitrogen was 
evaluated at three a da torn placements as opposed to one, and the hit, 
no-hit cases were examined separately. 



49 



B. MOLYBDENUM RESULTS 

1 , Nitrogen S p u 1 t e r i n g c r o & 5 s e c 1. 1 n s a n cj a cl a t om P I a c erne n t , 
a. 500 ev ion Bombardment 

The sputtering of nitrogen as a function of adatom place- 
ment was examined at two energy levels, 500 ev and 2 KeV. In addition, 
the "hit" and "no-hit" cases were examined separately. Table 10 lists 
the nitrogen sputter yield, and cross section as a function of adatom 
Placement for the 500 ev bombardment. The results are plotted in Figure 
11 for the hit, no-hit and total cross sections. The plot of cross sec- 
tion vs atom placement results in an apparent "well" corresponding to an 
adatom placement at 0,245 A above the (100) surface. This was true for 
both the hit and total cross section cases. The sputtering yield then 
increased as a function of adatom distance from the surface. The simula- 
tion cross section at this point is 5.63 * io" 1 ^ cm 2 , corresponding to 
the experimental value obtained by Winters of 5,6 x 10 ~' 6 cm 2 . 

The effect of the ion hitting the nitrogen is quite 
apparent. The sputtering cross section of the hit case increased by a 
factor of from 2 to 4 above the non-hit case, and by a factor of about 
1,5 for the total cross section. This indicates that the primary mecha- 
nism for the sputtering of the nitrogen is when the incident ion hits 
the adatom, and it in turn is reflected from the substrate, as illus- 
trated earlier in Figure 2(a). This seems to be more pronounced as the 
adatom is placed further above the "well" at .245 a. The agreement 
between the experimental and simulation results for this point indicates 
that the nitrogen may be located at approximately that height above the 
(001) surface, The nitrogen yields and cross sections are listed below 



50 



NITROGEN SPUTTER CROSS SECTIONS 



o 


OVJU A-j » i\iv\uui/ V-/*! *>i.vy i 















o 


LEGEND 


— 


a TOTAL CROSS SEC. 


/ 

/ 
i 
/ 




o 


o "HIT" CROSS SEC. 


in- 


a "NO-HIT" CROSS SEC. 


/ 

/ 
/ 
i 




o 


THEORETICAL VALUE: 6.0 E-16 


■*• - 


EXPERIMENTAL: 5.6 E-16 


i 
i 




o 




ri - 


of' 


CM) 

IL'.O 


i 
i 
i 
i 

i 






X E-16 SQ 

9.0 10.0 It 


i 
<> * 






2 o 

c d- 

O c 


q / 










o 












© 




— -A... 




O 








n* 








o 








0J~ 








o 








-" 


■ 






o 








o_ 




_. ... 





-0.1 



0.0 0.1 0.2 0.3 0.4 

ADAT0M PLACEMENT (ANGSTROMS) 



0.5 



F igure 1 1 . 



Comparison of "Hit", "No-hit" and Total Cross. Sections 
With Theory and Experimental values for 500 ev Ar on 
hoi ybdenum. 



51 



with H corresponding to the "hit" case, N for the "no-hit" case, and r 
for the total values. 

TABLE 10. NITROGEN SPUTTERING DATA, Mo (001), 500 eV 



Adatom 




Sputter 


Cross 


P I a cement 




Yield 


Sect ion 


(A) 






(x 10" 16 cm 2 ) 




H 


0. 41 


9. 1 


-0.05 


N 


0. 15 


4.3 




T 


0.28 


6.7 




H 


0.52 


10.5 


0. 146 


N 


0. 13 


2.68 




T 


0.33 


6.57 




H 


0.42 


8.47 


0.245 


N 


0. 14 


2.81 




T 


0.28 


5.63 




H 


0.63 


12.8 


0.290 


N 


0.20 


4.02 




T 


0.42 


8.4 




H 


0.66 


13.3 


0.335 


N 


0.20 


4.02 




T 




8.67 




H 


0.81 


16.3 


0.380 


N 


0. 19 


3.75 




T 


0.50 


10.02 



b. 2 keV Bombardment 

Nitrogen sputtering when bombarded at 2 keV was examined 
at the same adatom locations, and the data is tabulated in Table 11. 
Figure 12 shows the nitrogen sputtering cross sections as a function of 
adatom location, with the hit and no-hit case considered at four loca- 
tions. The curves follow the same general trends as Figure 11, with some 
exceptions. The same "well" is observed, but in this case, the total 



52 



NITROGEN SPUTTER CROSS SECTIONS 



o 




2.0 KEV AR<001> ON M0(001) 










i 

i 

i 
i 

i 




o 

o 

1ft- 

o 


LEGEND 
a TOTAL CROSS SEC 
o "HIT" CROSS SEC 
a "NO-HIT" CROSS SEC 

CALC: 8.0 E-16 

EXPERIMENT: 6.8 E-16 


© 






ri- 






o 

r > 
w O 

w 

cc °. 

~H o - 

1 - 
M " 1 o 
X o»- 




Q 

\ 
\ 
\ 
S 
\ 
\ 

\ 
\ 

\ 
\ 


Sv o / 




2 o 




^k 


/ . . 




^ S3 

b o 




CROSS S 

4.0 5.0 CO 

i i i 






.A 
A'' 




o 

CO" 




A... 


/ 




o 
cvi - 






"'■& 




© 










© 
d 




1 


1 i 





-0.1 



0.0 0.1 0.2 0.3 0.4 

ADAT0M PLACEMENT (ANGSTROMS) 



0.5 



F^ure 12. comparison of "Hit", -No-hit- and Total Cross Sections 

With Theory and Experimental values for 2 k*v Ar on 

Mo I yDdenum. 



53 



cross section of 5.6 x 10~ Uj cnr- is below the experimental and theoreti- 
cal values of 6.8 x 10 -1& cm 2 and 8.0 x 10" 1& cm 2 respectively. The 
slope of the sputtering cross section curve seems to level out at adatom 
placements above 0.29 A. The cross section for the -0.05 placement is 
below the "well" at 0.245 A. The -0.05 point was plotted using Webb's 
data. Both he and Meyerhoff used 14 N in their computations, where 15 N 
was used in this study, and m the experiments by Winters. The IV. mass 
difference may account for part of the difference in the result. The 
effect of the hit case is again quite apparent, with significantly 
higher sputter cross sections resulting from the ion striking the 
nitrogen atom. 

TABLE 11. Mo-N SPUTTERING CROSS SECTIONS, 2.0 KeV 

Adatom Sputter Sputter 

Placement Yield Cross Section 

(A) (X 10 -16 cm 2 ) 

-0.05 T 0.26 5.2 

H 0.56 11.2 

0.146 N 0.15 3.0 

T 0.35 7.01 

H 0.45 9.02 

0.245 N 0.11 2.2 

T 0.28 5.6 

H 0.79 15.9 

0.290 N 0.20 4.0 

T 0.50 10.03 

H 0.80 16.1 

0.335 N 0.23 4.60 

T 0.51 10.3 

0.380 T 0.53 10.6 



54 



c • C ornp a r i s o n o f t h e s p utter • n g c r o s s S e c t > o n g 

The total sputtering cross sections for 500 ev and 2 KeV 
are plotted on the same graph in Figure 13. The general trends are 
apparent. Both curves exhibit the same "welt" corresponding to an adatom 
Placement of 0.245 a. The sputtering cross sections then increase as a 
function of increasing height of adatom placement. There seems to he 
something special about the 0,245 A placement. Since the simulation 
cross section agreed so closely with the experiment, it is a good indi- 
cation that this might be close to the actual location of the atom. The 
fact that the 2 KeV curve so closely matches the 500 eV curve gives fur- 
ther credence to this assumption, 
2 . Nitrogen Sputtering Cross Sections as a Function of Energy 

The sputtering of nitrogen as a function of energy was 
examined. The purpose of this analysis was to compare the results of 
the simulation to the data presented by Winters as listed in Table 9. 
Points were obtained at four energy levels, 500 eV and 1.0, 2.0, and 3,0 
KeV. The analysis was conducted for adatom placement at 0.245 A. This 
point was chosen, since it so closely matched the experimental cross 
section as mentioned above. The results are shown in Figure 14, with 
the corresponding experimental and theoretical values from Winters' 
paper, The values at 500 ev and 3.0 Kev are within the bounds of the 
experimental and theoretical values, but the points at 1 KeV and 2 Kev 
ar^ beyond the limits. The variation of the simulation results from 
those of Winters is a measure of the uncertainty of the simulation 
model. However, the general trend of the simulation cross sections do 
correlate with Winters' results. 



55 



o 

o 
to 

o 



o 



o 



NITROGEN SPUTTER CROSS SECTIONS 

.5 AND 2 KEV AR<001> ON M0(001) 



c- 



JVJ 



O C3- 

CJ o 
Id ;v 

CO 

O 
K o 



O 



o 

C*5 



o 



o 
o. 



LEGEND 
□ TOTAL CROSS SEC: 500 EV 
o TOTAL CROSS SEC: 2.0 KEV 



-o.i 




0.0 0.1 0.2 0.3 0.4 

ADATOM PLACEMENT (ANGSTROMS) 



0.5 



Figure 13, 



Comparison of Total Cross Sections for 500 eV and 2 kev Ar 
on Molybdenum. 



56 



NITROGEN SPUTTER CROSS SECTIONS 



o 




N AT 0.245 AN, MO(OOl) 












o 








o 


a 


LEGEND 
SIMULATION RESULTS 






o 

o 



A 


THEORETICAL VALUES 
EXPERIMENTAL RESULTS 


d- 








o 
O o 

w 

— i = - 
1 ~ 








H 






..■■• -° 


CROSS SECTION 

5.0 CO 7.0 0.0 

i i i i 








o 








© 
d~ 








o 
cvi" 








© 








o 
©_ 




1 1 1 ...... , , 





0.0 



0.5 



1.0 1.5 2.0 

ION ENERGY (KEV) 



2.5 



3.0 



3.5 



Figure 14. Nitrogen Sputtering Cross Section as a Function of Energy 
for Adatoms Placed at 0,£4? a on Mo (100), 



57 



3 . Spu tter t '"'■.) of ^ h «? Substrate 

The ratio of the reacted substrate yield to the bare substrate 
yield was examined at 500 ev and 2 Kev. The hit, no-hit and total 
yields were considered. The results are listed in Table 12, and are 
plotted in Figure 15. in all cases, for all a da torn placements examined, 
the yield of the reacted surface was lower than that of the thire sur- 
face. Additionally, the substrate yields of the hit case were lower 
than for the total and non-hit case. This indicates that the nitrogen 
adatorns absorb some of the energy of the incident particle, and there- 
fore less momentum is available to initiate the collision cascade in the 
substrate. 

The case was the same at 2 Kev. The sputtering yields of the 
substrate were lower in each case when the a da torn was initially struck 
by the ion. The incident ion at 2 Kev still failed to have sufficient 
energy to impart enough momentum to the a da torn for it to act as a second 
projectile and thereby enhance the sputter yield of the substrate. The 
yields of the substrate and the ratios with the bare substrate for 2 Kev 
are also listed in Table 12. 

C. TUNGSTEN RESULTS 

\ . Nitrogen sputtering Cross Sections and Adatorn Placement 
a. 500 ev ion Bombardment 

The sputtering of nitrogen from (001) tungsten was 
examined at 500 ev and 2 Kev. The hit and no-hit scenarios were 
examined as with molybdenum, The results are listed in Table 13, and 



58 



TABLE t£. SPUTTER YIELD OF SUBSTRATE, Mo (100) 



A da torn 

P l a cement 

(A) 



Yield 



500 ev 



W^b 



Yield 



2 kev 



y n /Yb 



Bare 



05 



0. 146 



0.245 



0, 290 



0,335 



0,380 



H 


3 


16 


N 


3 


41 


T 


3 


26 


H 


1 


99 


N 


2 


29 


T 


2 


14 


H 


1 


51 


N 


2 


37 



T 2 . 09 

H 1,91 

N 2.45 

T 2, 17 

H 2.03 

N 2.51 

T 2.27 

H 2.08 

N 2 . 60 

T 2 . 34 

H 2.03 

N 2 . 59 

T 2.31 



NA 






NA 






NA 


3.02 


NA 


0, 630 






0.671 






0.652 


2.29 


. 76 


0,573 


2.01 


0. 67 


0.695 


3.07 


1 .02 * 


0. 637 


2.54 


0,64 


0, 604 


1 .85 


0.61 


0.716 


2.66 


0.88 


0.661 


2.25 


0,74 


0,642 


2.26 


0.75 


0, 736 


3.15 


1 .04 * 


0.692 


2.72 


0,90 


0,658 


2,09 


0. 69 


0.762 


3, 17 


1 .05 » 


0.713 


£ . 63 


0.67 


0.642 






0.760 






0.704 


2.25 


0.74 



** it should be noted that the results for the hare substrate 
were taKen from reference 50. The 2 Rev Mo-bare case was not 
re-run. This could be part of the reason for the ratio of 
sputter yields for the new potential functions to be greater 
than 1. it does not seem reasonable that the yield of the 
reacted substrate should be greater than the bare substrate 
when the adatom is not struck. 



59 



SPUTTER YIELDS OF SUBSTRATE 



© 




REACTED/BARE SPUTTER YIELD, MO 




eo 






d ~ 




>v / ^,© V) 


in 




T „&' 


Q 




___-o-~' ^-e— -^__, 


1 




-o--""" ^^^^ B 


H 


o 




P 




5- ..A" "A 


g o 


A-.... 




5«- 




"' ••-.... ... .-A - ' 


< 




. — " 


a 




"-A*" 


o 






H 






Q 






Erf 






u-, 






CJ 






< 






fc3 -»• 






r-"* — >' - 






t— i o 






Ex. 






O 






o 






< 




LEGEND 
□ 500 EV TOTAL SPUTTER YIELD RATIO 




CJ 




o 500 EV "NO-HIT" YIELD RATIO 




d - 




"a"500~EV ""HIT"" YIELD RATIO 






+ 2 KEV TOTAL YIELD RATIO 


o 








d 




1 



-0.1 



0.0 0.1 0.2 0.3 0.4 

ADAT0M PLACEMENT (ANGSTROMS) 



0.5 



Figure 15, 



Ratio of the Sputter Yield of the Reacted Substrate to the 
Yield of the Bare Substrate, 



60 



are plotted in Figure 16. The same general pattern wa v Found as foi 
molybdenum. The sputtering cross section for the hit scenario is again 
significantly greater than for the no-hit case, for all adatom place- 
ments. Therefore, as with molybdenum, the majority of the nitrogen is 
sputtered as a result of a direct collision by the incident ion. The 
total sputtering cross sections generally follow the experimental val- 
ues until the pomt corresponding to the 0,2464 a adatom location, where 
the slope of the cross section plot rapidly increases. 

b. 2 KeV Ion Bombardment 

The sputtering cross section and yield data for the 2 kev 
bombardment is summarized in Table 14 and plotted in Figure it. There 
is no "well" as was observed for the molybdenum case, however there is a 
break in the slope of the curve at the pomt corresponding to adatom 
placement at 0.2464 A. While not plotted, the sputtering cross sections 
for the hit case were again significantly greater in all instances than 
for the non-h i t case, 

c . C ornp a r i s o n o f S p u 1 1 e r i n g C r o s s 8 e c 1 1 o n s 

The total nitrogen sputtering cross sections for the 500 
eV and 2 keV bombardments art plotted in Figure 19, The two curves fol- 
low the same general trends. A "well" is not observed, but there is an 
inflection point in the curves at a point corresponding to an adatom 
placement of 0.2464 A. One special note, the sputtering cross section 



61 



o 



c 



o 



q 



NITROGEN SPUTTER CROSS SECTIONS 

500 EV AR<001> ON W(001) 



U 

C/2 - 
O 

I o 

C 9 

a 
w 

o 
OS 

u 

o 



© 



© 

d. 



LEGEND 
□ TOTAL CROSS SEC 500 EV 
o "HIT" CROSS SECTION 
A"' 7r N0-HTT ,f CROSS'SECTION 

500 EV EXPER: 0.6 E-16 

500 EV THEORY: 6.9 E-16 



,-Q 



..A' 



0' 



-0.1 



0.0 0.1 0.2 0.3 0.4 

ADAT0M PLACEMENT (ANGSTROMS) 



0.5 



Figure 16. Comparison of Hit, No-hit and Total Nitrogen Sputtering 
Cross beet ion with Experiment and Theory for <soo ev Argon 
on w(ioo) J 



62 



o 



NITROGEN SPUTTER CROSS SECTIONS 

2 KEV AR<001> ON W(OOl) 



O 
CO 



O 

in 



o 



o 



o 

© 

cr 
o 



o 
d. 



LEGEND 
g TOTAL CROSS SEC: 2.0 KEV 

2.0 KEV EXPER: 9.0 E-16 
'2.0 KEV'THEORY: 8.1E-16" 




-o.i 



0.0 0.1 0.2 0.3 0.4 

ADAT0M PLACEMENT (ANGSTROMS) 



0.5 



Figure 17. Nitrogen Sputtering Cross Section as d Function of Adatom 
Placement for 2 keV Argon on W{100). 



63 



NITROGEN SPUTTER CROSS SECTIONS 



o 


• o m>i/ k* ivjl^i * n.i^^v-'vj 


X^ \-> 1 1 II \\J <J i ) 












o 








CO- 

ml 








O 








Ift- 








© 

o 


LEGEND 
□ TOTAL CROSS SEC 500 EV 
o TOTAL CROSS! SEC: 2.0 KKV 






ri - 

m4 








o 


. 










-1G SQ 

10.0 11 








W 






„-- e 


Z o 

C ss- 

C/3 








CROSS 

5.0 CO 

i i 


G'"' 






© 








•*~ 1 








o 








ci- 


. 






ts 








CJ~ 








© 








-*~ 








o 
d 


1 . . 




1 



-0.1 



0.0 0.1 0.2 0.3 0.4 

ADAT0M PLACEMENT (ANGSTROMS) 



0.5 



F i gure 18. 



Comparison of Total Nitrogen Sputtering Cross 
Sections for 500 eV and 2 keV Argon on W(100) 



64 



TABLE 13. WMOOi NITROGEN SPUTTERING RESULTS, 500 eV 



A da torn 
Placement 

(A) 



-,05 



0, 1623 



0.2464 



0.325 



0.487 



Sputter 


Sputter 


Y i e l d 


C r o a s S e c t i o n 




(x 10 -16 cm 2 ) 


H ■ 60 


12.3 


N 0,23 


4.68 


T 0,42 


8 . 56 


H 0,69 


14.1 


N 0.23 


4.68 


T 0.46 


9,36 


H , 68 


13.9 


N 0.27 


5.49 


T 0.47 


9,59 


H 0.83 


17,0 


N 0.32 


6.51 


T 0,57 


11,6 


H 0.99 


20.3 


N 0.33 


6.72 


T 0,66 


13,5 



TABLE 14, NITROGEN YIELD AND SPUTTERING CROSS SECTIONS FROM TUNGSTEN 
2,0 kev 

A da torn To t a l spu t ter 

Placement Sputter Cross Section 

(A) Yield (x 10~ 16 cm 2 ) 



05 0.28 5,67 

1623 0,35 7.08 

2464 0,36 7.28 

325 0,40 8,09 

487 0,45 9.11 



65 



of nitrogen at 500 eV i s & 1901 f icanti y greater than ai 2 kev for j 1 i 
adatom placements. This is similar to t ne findings in the two prev> 
theses [Ref. 49, 50], This conflicts with the experimental and theoret- 
ical values presented by Winters. This is an indication that the model 
used in the simulation is not correctly portraying the behavior of the 
n 1 trogen- tungsten system. 

2 . Nitrogen Sputtering as a Function of Energy 

in an effort to find an adatom placement location that agreed 
with the experimental results, the total nitrogen sputtering cross sec- 
tion was plotted as a function of energy for all five adatom locations 
tested. The results are shown in Figure 19, In all cases, there was a 
strong negative slope, that did not correspond well with the slight pos- 
itive slope of the experimental and theoretical values. Again this cor- 
responds to Webb's and Meyerhoff's findings. This indicates that the 
potential function may not be reflecting the true behavior of the 
nitrogen on the tungsten. 

3. sputtering of the Substrate 

The analysis of the ratio of the sputtering yield of the 
reacted substrate to the bare substrate was not conducted to the same 
detail as with mo l ybdenum. The results for the cases examined are sum- 
marized below m Table 15, in all cases, with the exception of the hit 
case for placement below the surface at 500 eV, the yield of the reacted 
substrate was lower than that of the bare substrate, This was the Cdse 
both at 500 eV and 2 KeV. This indicates that for these two levels, the 
adatoms again reduce the energy of the ion such that less momentum is 
transferred to the substrate, and the yield of the substrate is lowered. 



66 



a 

r^ 

*—■ 

c 

50 

o 

in- 

o 

o 
ci ■ 

o 

c:- 
c/: 



w 

X ci 



NITROGEN SPUTTER CROSS SECTIONS 

AS A FUNCTION OF ENERGY, W(001) 



o 

CO 

GO 
CO 

o 



o 



o 

© 

© 

ev 

o 
ci 



© 

d. 



a 


LEGEND 
EXPERIMENTAL VALUES 





CALCULATED VALUE 

N AT -.05 ANGSTROMS" 


+ 


N AT .1623 ANGSTROMS 


X 


N AT .2164 ANGSTROMS 


O 


N AT .325 ANGSTROMS 


V 


N AT .407 ANGSTROMS 






0.0 



0.5 



1.0 1.5 2.0 

ION ENERGY (KEV) 



2.5 



3.0 



Figure 19. Comparison of Nitrogen Sputtering Cross sections as 
a Function of Energy and Adatom Location, W(100). 



67 



One interesting observation is That the yield of the substrate at 2 Kev 
d i ("J increase, as expected, at the higher energy, This despite the fact 
that the simulation cross section of nitrogen sputtering was higher at 
500 ev than at 2 kev. 

TABLE 15. SPUTTER YIELD OF SUBSTRATE W(IOO) 



500 ev 



2 kev 



A da torn 








P la cement 




Yield 


WY B 


A 










H 


2,67 




Bare 


N 


3.25 


NA 




T 


2.98 






H 


1 .79 


0. 67 


-0.05 


N 


1 .75 


0,53 




T 


1 .77 


0.59 




H 


1 .83 


0,66 


0. 1623 


N 


1 .68 


0.57 




T 


1 .85 


0,62 




H 


1 .77 


0. 67 


0,2464 


N 


1 ,9 


0.57 




T 


1 .84 


0.62 




H 


1 .76 


0. 66 


0,325 


N 


2.05 


0.63 




T 


1 .90 


0.64 




H 


1 .73 


0.65 


0,467 


N 


2.23 


0.66 




T 


1 .98 


0,66 



Yield 



2 . 63 

2. 16 
1 ,57 
1 ,67 



2.06 



2.35 



50 



1 .96 



Y N/ y B 



NA 

0.82 
0,60 
0.71 



0.78 



0.69 



0,95 



0.74 



D. COMPARISON OF MOLYBDENUM AND TUNGSTEN RESULTS 
1 , Comparison of Cross Sections 

The sputtering cross sections of nitrogen from moi ybdenum and 
tungsten are plotted in Figure 20, with the adatom placement normalized 
to lattice units. The molybdenum curves exhibit the "well" at 0.156 LU, 



68 



o 

c 
d 

o 



c 






NITROGEN SPUTTER CROSS SECTIONS 

.5 AND 2KEV AR ON W AND MO(OOl) 



<2 

CO ° 



a - 



>S d 

O a 

CJ o 

CO 
CO o 

in 6 

c 
^^ 



c 
o 

o 

CJ 



o 

c 







LEGEND 
□ 500 EV AR<001> ON M0(001) 




o 2.0 KEV AR<001> ON M0(001) 
a 500 EV AR<001> ON W(001) 


- + 2.0 KEV AR<00i> ON W(00i) 




..&■+ 7 

p— — ,•••'■"" ,'•■ 7 


r'' 


1 i 





-0.05 0.00 



0.05 0.10 0.15 0.20 0.25 

ADAT0M PLACEMENT (LU) 



0.30 



0.35 



Figure 20. Comparison of Nitrogen Sputtering Cross Sections 

for Mo (100) and W(100) at 500 eV and 2 keV, Adatom 
Placement Normalized to Lattice Units. 



69 






but the tungsten curves show a definite "breaK" in slope &x the -j r r •»— 
relative point. This point corresponds to 0.245 a for molybdenum, and 
0.2464 a for tungsten. The potential function for the 0.2464 LU place- 
ment was derived so that the behavior at the point corresponding to the 
molybdenum well could be examined. Again the major difference between 
the two substrates is the fact that the sputtering cross section at 500 
ev is higher than at 2 keV for tungsten, 

The hit and no-hit analysis was conducted for tungsten also, 
and the same general trend was noted. Significantly higher sputtering 
cross sections were noted for the hit scenario. This was true at all 
energy levels and adatom placements examined. This indicates that the 
primary mechanism for the sputtering of nitrogen is due to direct colli- 
sion by the incident ion, 

2. Comparison of Substrate Sputtering Yields 

The ratios of the sputtering yields of the reacted substrates 
to the bar^ substrates at 500 ev and 2 kev bombardments indicate that 
the nitrogen does not enhance the sputtering of the substrate. This is 
observed even at the relatively high energy level of 2 kev. Both 
molybdenum and tungsten exhibited this behavior. The incident ion, even 
at 2 kev failed to transfer enough momentum to the nitrogen for it to 
act as a second projectile to enhance the sputtering of the substrate. 

E. DETERMINATION OF POSSIBLE MASS EFFECTS 

One of the major conclusions in Winters' earlier works was the fact 
that the sputtering cross section seemed to be greater for the substrate 
with the higher mass [Ref. 48, 55], Meyerhoff and Webb examined this 



70 



aspect, and concluded that adatom placement ftdO a greater effect, rhe 
ma s s effect wa s e x am i ned in t h i s thesis by s e 1 e c t i ng t h r ee a d a t om I o c .1 - 
tions in the vicinity of the "well" at 0,245 a, for the Mo (100) surface, 
and replacing the mass of the molybdenum with the tungsten mass, leaving 
all other aspects of the potential functions the same. The results are 
plotted in Figure £1, for the hit, no-hit and total cross section cases. 
In all cases, the sputtering cross sections were higher for the higher 
mass. This indicates that the mass of the substrate does have an effect 
on the sputtering cross section of the nitrogen. 

Figure 22 shows the total nitrogen sputtering cross sections for 
500 ev Mo(100), Mo*MOO) (the high-mass molybdenum), and W(100), The 
graph is normalized for adatom placements at approximately corresponding 
locations. The trend for higher sputter cross sections for the higher 
mass is obvious. The heavy Mo* has higher nitrogen sputtering cross sec- 
tions than the true molybdenum, and the tungsten has the highest cross 
sections of the three. Through the use of the computer simulation, the 
mass effects n^y^ been able to be truly isolated, more so than is pos- 
sible in a physical system. This is one valuable aspect of simulations. 

One possible reason for the higher sputtering cross section for the 
heavier substrate is that the greater mass of the substrate provides a 
better "springboard" for the nitrogen atoms. When the nitrogen atoms 
are KnocKed from their positions by the incident ions, they relatively 
light nitrogen atoms will rebound with more initial momentum conserved 
from the heavier tungsten atoms than from the lighter molybdenum atoms. 

This model for the increase in sputtering yield with higher sub- 
strate mass follows if the primary mechanism for the sputtering of the 



71 



NITROGEN SPUTTER CROSS SECTIONS 

500 EV AR OX MO AND M0*(001) 



O 



o 



o 
to. 



LEGEND 

□ TOTAL CROSS SEC. 

o "HIT" CROSS SEC. 
r" "NO-HIT" CROSS SEC. " 
+ MO* "HIT" CROSS SECTION 
x" " M0 V "NO- HIT' r CROSS'SECV 
o"" MO* TOTAL CROSS SEC." 



.-H-- 



U 

c 
to c 

O cvi- 

T. 

w 

><o 

o 

CJ 

c/: =' 

CO 
CO 

K o 
CJ cd- 



©-. 




© 
ci' 



X" 



.-A 



o 

d. 



0.10 



0.15 0.20 0.25 

ADAT0M PLACEMENT (ANGSTROMS) 



0.30 



F igure 21 . 



Comparison of Nitrogen Sputtering Yields for 
Mo (100) and Mo* (100), 500 eV Argon. 



72 



o 

o 
se . 

© 

d ■ 



O 

d ■ 



o 



NITROGEN SPUTTER CROSS SECTIONS 

500 EV AR ON MO, MO*, AND ff(001) 



O q 

w 

I - 

Z o 

O d- 

U o 
W 



o 



o 
d' 

o 

© 

o 
d' 

o 
c\i" 



o 
d. 



LEGEND 
MO TOTAL CROSS SECTION 



o MO* TOTAL CROSS SEC. 
a W TOTAL CROSS SEC 



0.05 




0.10 0.15 0.20 

ADAT0M PLACEMENT (LU) 



0.25 



F igure 22. 



Comparison of Total Nitrogen Sputtering Cross 
Sections for Mo(100), Mo*(100), andW(100), 500 eV 
Argon. 



73 



n 1 1 1" o g e f"i i s <:! i r e c t c oil i 8 i o r i by the i n c i d en t i o n . if The nit r o 9 e n a t oms 
were KnocKed off by reflected argon ions, or by sputtering substrate 
materials to a high degree, the effect of the mass of the substrate 
would seem to be less important. 



F. COMPARISON WITH PREVIOUS SIMULATIONS 

The results of this study indicates that there is a significant 
mass effect, this is contrary to the findings of webb and Meyerhoff. 
The earlier studies did not isolate the effects of the direct interac- 
tion of the 1 on with the adatom to the degree that it was done in this 
study. The trend for the higher nitrogen sputtering cross section at 
the lower energy for tungsten is consistent through the three studies. 

The other aspects of the computer simulation, such as the angular 
and energy distribution plots remained consistent with earlier works. 
For a more detailed analysis of the "traditional" sputtering simulation 
analysis, the reader is referred to References 49 and 50. 



74 



V. CONCLUSIONS AND RECOMMENDAT I ONS 

The results of this simulation confirm the findings of Webb and 
lieyerhoff that the placement of the adatoms is most likely above the 
crystal surface. A location of approximately 0.245 A for the molybdenum 
surface results in a nitrogen sputtering cross section that agrees with 
the experimental results of Winters. 

Comparison of the "hit" and "no-hit" scenarios indicates that the 
primary mechanism for the sputtering of the adatom is due to direct 
collisions with the incident ion. In all cases examined, the sputtering 
cross sections derived when the adatom was struck by the incident ion 
were larger than in the cases when they were not struck. This was found 
at both energy levels examined, 500 eV and 2 keV; and for both Mo (100) 
and W(100). This agrees partially with Winter's conclusions, but does 
not show a significant increase in non-hit sputtering at higher energies 
he proposed. 

The mass effect of the substrate was examined, and it was deter- 
mined that the sputtering cross section of the adatom was enhanced with 
the increase of mass of the substrate, when all other factors are held 
constant. This contradicts Meyerhoff's findings, but only one point was 
considered in his analysis. 

The ratio of the sputtering yield of the bare substrate to the 
reacted substrate was evaluated for both systems. The analyses indi- 
cated that the adatoms decrease the amount of energy imparted to the 



75 



substrate, and reduces the sputter yield for the "hit" ca*e. Fh i 
momentum loss was observed dt 500 eV and 2 keV. This process c<>>id 
change at higher energy levels (greater than 2 keV) of the incident par- 
ticles. 

A significant difference was found for tungsten between the simula- 
tion and experimental and theoretical sputtering cross sections. The 
nitrogen sputtering cross section for the 500 eV bombarding energy was 
higher than for the 2 kev energy, which was in contradiction to Winter's 
results. A similar finding was made by Webb. This indicates that per- 
haps the pair potentials used to model the nitrogen on tungsten are not 
correct. It has been determined that the oxygen-tungsten system behaves 
quite differently than the oxygen-molybdenum system, so perhaps this 
trend continues for the Mo-N and W-N systems [Ref. 67], 

Further simulation studies should be made to complete the compari- 
son with Winters' study. Studies should be made using helium and xenon 
ions m order to investigate the effect of the mass of the incident ion. 
Additional studies should be conducted on tungsten to determine if the 
lower sputtering cross section at 2 keV was due to nitrogen atoms being 
captured as i nterst i t ia Is in the tungsten lattice, or if there is a 
defect in the potential functions. Higher energy studies should be 
conducted to identify the point at which the sputtering yield of the 
substrate will be enhanced by the adatoms. 



76 



APPENDIX 
DERIVATION OF SPUTTERING CROSS SECTIONS 

The methodology used for the derivation of the sputtering cross 
section m this thesis is based on the technique used by Winters m 
reference 55. The notation used in this appendix closely parallels that 
of Wi nter s . 

The rate that nitrogen is sputtered from the surface of the target 
is given by the relation: 



de N 

R N = = - ° e N v + t , ( 7 ) 

dt 



where the following are defined: 

R N = the nitrogen sputtering rate, {atoms / cm 2 - sec]; 

8n = the nitrogen atom concentration at time = t, {atoms / cm 2 !; 

a s the sputtering cross section, {cm 2 ]; 
V + h the incident ion flux, {ion / cm 2 - sec]; 
t = the time in seconds, {sec}; and 
V + t = the ion dose, (fluence), {ion / cm 2 ]. 



Solving equation 7 yields, 

In (G N / No ) = -oV + t, ( 8 ) 

where 9^ = the initial nitrogen concentration at t = 0. Rearranging 
equation 8, and solving for a results in the following relation for the 
cross section; 

a z -in O N /e N0 ) / V + t . ( 9 ) 



77 



The ©n/©n term ,s next examined. The following relations are 
def i ned, 

No = the number of nitrogen atoms on the surface at t = 0, 

NTRAJ = the number of trajectories run, also 

NTRAJ = the number of incident ions on the target, 

NSPUTT = the number of nitrogen atoms sputtered from the surface, 

A = the area of the target {cm 2 ). 

Consequently, for each trajectory we have 
N = (N - NSPUTT) / A , 



and 9 No = N /A . Thus, 



9k 



(N - NSPUTT) / A 



(N - NSPUTT) 



e 



No 



N / A 



Summing over all trajectories, 



0K 



e 



NO 



NTRAJ 



i - 1 



(N - NSPUTT) j 



1 - 



NTRAJ 
Y NSPUTT 



i = 1 



(N ) (NTRAJ) 



NTRAJ 
f we define £ NSPUTT = TOTSPUT, then 

i - 1 



78 



6k 



1 - 



e 



NO 



TOTSPUT 



(N ) (NTRAJ) 



( 10 ) 



Incorporating the relation for the beam fluence, from equation 9 
yields the following relation for the cross section; 



a = 



■( 



TOTSPUTT 



(N ) (NTRAJ) 



TOTSPUTT 

but = YIELD, so as a result; 

NTRAJ 



A ■ 



o - 



(yield\ 



( 11 ) 



Equation 11 is the relation used to calculate the sputtering cross 
section for the adsorbed nitrogen from the tungsten and molybdenum sur- 
faces. 



79 



LIST OF REFERENCES 



1. Grove, W.R., "On the Electro-Chemical Polarity of Gases," Trans . 
Ro ya I Soc lety of London , v. 142, pp. 87-102, 1852. 

2. Plucker, J., "For tgesetzte Beobachtungen Uber die Elekthische End- 
ladung durch Gasverdunnte Raume," Anna len der Phys i K , v. 104, 
p. 113, 1858. 

3. Plucker, J., " Ueber die Emwirkung des Magneten auf die Elektri- 
schen Entlandungen in Verdunnten Gasen," Annalen der Physik , 

v. 103, p. 88, 1858. 

4. Plucker, J., "Fortgestezte Beobachtungen Uber die Elektrische Ent- 
ladung," Annalen der Physik , v. 105, p. 67, 1858. 

5. Gassiot, J. P., "On the Stratifications and Dark Bands in Electrical 
Discharges as Observed in Torricellian Vacua," Trans. Royal Society 
of London , v. 148, pp. 1-16, 1858. 

6. Goldstein, E., "The Canal-Ray Group," Verh. Dtsch. Phys. Ges. , 
v. 4, pp. 228-237, 1902. 

7. Stark, J., "Volatilisation by Atom Rays," Zeitschrift fur Elektro- 
chem. , v. 15, pp. 509-512, 1909. 

8. Thompson, J.J., Rays of Positive Electricity and Their Applications 
to Chemical Analyses , Longmans, Green and Co., 1921. 

9. Bush, V. and Smith, C.G., "Control of Gaseous Conduction," Trans. 
of Amer . Inst, of Electrical Engineers , v. 41 pp. 402-411, 1922. 



10. 



1 1 



12 



Kmgdon, K.H, and Langmuir, I., "The Removal of Thorium from the 
Surface of a Thoriated Tungsten Filament by Positive Ion Bombard- 
ment, " PJiy_£j_cjJ_Rev_ilw, v. 22, pp. 148-160, 1923. 

Kingdon, K. H. and Langmuir, I., "The Removal of Thorium from the 
Surface of a Thoriated Tungsten Filament by Bombardment with Posi- 
tive Ions," Physical Review , v. 20, p. 108, 1922. 

Blechschmidt, E., "Die Kathodenzer staubung in Abhangigkeit von den 
Betr i ebsbedingungen, " Anna len Der Phys i k , v. 81, pp. 999-1042, 
1926. 



13. Blechschmidt, E. and von Hippie, A., "Der Einfluss von Material und 
Zustand der Kathode auf den Zer staubungsprozess , " Annalen der 
Phys ik , v. 86, pp. 1006-1024, 1928. 



80 



14. von Hippie, A., "Uber die Nat lit unci den Ladungszustand der bei 

kathodenzer staubung Emittierten Meta I I te i I chen, " Anna ten Der Phy- 
5 i K , v. 80, pp. 672-706, 1926. 

15. Von Hippie, A., "Zur Theorie der Kathodenzerstaubung, " Anna 1 en Der 
Phys iK , v. 81, pp. 1043-1075, 1926. 

16. Lamar, E.S. and Compton, K.T., "A Special Theory of Cathode Sput- 
tering," So lence , v. 80, p. 541, 1934. 

17. Penning, F.M. and Moubis, J.H.A., "Cathode Sputtering in a Mag- 
netic Field," Konmkl. Ned. Akad. Wehenschap. Proc, , v. 43, pp. 41- 
56, 1940. 

18. Kewell, F., "A Mechanism for Sputtering in the High Vacuum Based on 
the Theory of Neutron Cooling," Physical Review , v. 87, pp. 160- 
161, 1952. 

19. Wehner , G.K., "Momentum Transfer in Sputtering by Ion Bombard- 
ment," Journal of Applied Physics , v. 25, pp. 270-271, 1954. 

20. Wehner, O.K., "Sputtering of Metal Single Crystals by Ion Bombard- 
ment," Journal of Applied Physics , v. 26, pp. 1056-1057, 1955. 

21. Wehner, G.K., "Controlled Sputtering of Metals by Low Energy Hg 

Ions," Physical Review , v. 102, pp. 690-704, 1956. 

22. Harrison, D.E., Jr., "Theory of the Sputtering Process," Phys i ca I 
Review , v. 102, pp. 1473-1480, 1956. 

23. Behrisch, R., "Introduction and Overview," Sputtering by Particle 
Bombardment I , R. Behrisch, ed. , pp. 1-8, Springer, 1981. 

24. Townsend, P.D., Kelly, J.C., and Hartley, N.E.W., Ion Implanta- 
tion, Sputtering and Their Applications , Academic Press, 1976. 

25. Robinson, M.T., "Theoretical Aspects of Monocrystal Sputtering," 
Sputtering by Particle Bombardment I , R. Behrisch, ed. , Springer, 
1981 . 

26. Garrison, B.J., Winograd, N. , and Harrison, D.E., Jr., "Atomic 
and Molecular Ejection from Ion-Bombarded Reacted S mgle-Cr ystd l 
Surfaces. Oxygen on Copper (100) , " Phys i ca I Rev i ew B , v. 18, 

pp. 6000-6010, 1978. 

27. Winters, H.F. and Sigmund, P., "Sputtering of Chemisorbed Gas 
(Nitrogen on Tungsten) by Low Energy Ions," Journa I of App l i ed 
Phys ics , v. 45, pp. 4760-4766, 1974. 

28. Wood, E.A., "Vocabulary of Surface Crystallography," Journal of 
App l ied Phys ics , v. 35, pp. 1306-1312, 1964. 



81 



29. Carter, G. and Col I igon, J.S., Ion Bombardme nt of c ."l ids, *mer i in 
Elsevier Publishing Co., Inc., 1968. 

30. Sigmund, P., "Sputtering by Particle Bombardment: Theoretical Con- 
cepts," Sputtering by Particle Bombardment I , R. Behrisch ed. , 
Springer, pp. 9-71, 1981. 

31. Wehner , O.K., "Sputtering Yield Data in the 100-600eV Energy 
Range," General Hills Report 2309 , 1962. 

32. Wehner, G.K., Anderson, G.S., and KenKrnght, C.E., L i tton I ndus- 
tr i es Report 3031 , Litton Systems, Inc., 1966. 

33. Silsbee, R.H., "Focusing in Collision Problems in Solids," Journa 1 
of Appl ied Phys ics , v. 28, pp. 1246-1250, 1957. 

34. Sigmund, P., "Theory of Sputtering I. Sputtering Yield of Amor- 
phous and Po 1 ycr ysta 1 I i ne Targets," Phys i ca 1 Rev i ew , v. 184, 

pp. 383-415, 1969. 

35. Thompson, M.W. , "The Energy Spectrum of Ejected Atoms During High- 
Energy Sputtering of Gold," Philosophical Magazine , v. 18, pp. 377- 
414, 1968. 



36. Harrison, D.E., Jr., "Sputtering Models - A Synoptic View," Rad ia- 
tton Effects, v. 70, pp. 1-64, 1963. 



37. Kelly, R., "The Mechanisms of Sputtering. Part I. Prompt and Slow 
Collisional Sputtering," Radiation Effects , v. 80, pp. 273-317, 
1984. 

38. Gibson, J.B., and others, "Dynamics of Radiation Damage," Phys ica I 
Rev lew , v. 120, pp. 1229-1253, 1960. 

39. Robinson, M.T. and Torrens, I.M., "Computer Simulation of Ato- 
mic-Displacement Cascades in Solids in the Binary Collision Approx- 
imation," Phys ical Review B , v. 9B, pp. 5008-5024, 1974. 

40. Webb, R.P. and Harrison, D.E., Jr., "Near-Threshold Sputtering 
Mechanisms from a Computer Simulation of Argon-Bombarded Clean and 
Oxygen-Reacted Copper Single Crystals," Journal of Applied Physics , 
v. 53, pp. 5243-5249, 1982. 

41. Thompson, J.J. and Thompson, G.P., Conduction of Electricity 
Through Gases , third edition, Cambridge, 1928. 

42. Loeb, L.B., Fundamental Processes of Electrical Discharge in Gases , 
Wi ley, 1939. 

43. Taglauer, E., "Atomic Collision Processes During Plasma-Wall Inter- 
action in Fusion Devices," Nuclear Instruments and Methods in Phys- 
ics Research, v. B13, pp. 218-224, 1986. 



82 



44. Johnson, L.F., "Ion Beam Microstructure Fabrication in Optical, 
Magnetic and Surface Acoustical Technologies," Ion Bombardment 
Modification of Surfaces , 0. Auciello and R. Kelly, eds., Elsi- 
vier, pp. 361-397, 1984. 

45. Banks, B.A., "Ion Bombardment Modification of Surfaces in Biomedi- 
cal Applications," Ion Bombardment Modification of Surfaces , 0. 
Auciello and R. Kelly, eds. Elsivier, pp. 399-434, 1964. 

46. NASA, Sputtering and Ion Plating , A Conference Held at Lewis 
Research Center, Mar 16, 1972, NASA, 1972. 

47. Alnot, P. and King, D.A., "Trapping, Sticking, and Reactive Scat- 
tering in Chemisorpt ion: Nitrogen Isotopes on W(100)," Surface 

Sc lence , v. 126, pp. 359-367, 1983. 

48. Winters, H.F., "Mass Effect m the Physical Sputtering of Multi- 
component Materials," Journal of Vacuum Science and Technology , 
v. 20, pp. 493-497, 1982. 

49. Meyerhoff , 0. , Computer Simulation Studies of Sputtering From Clean 
Tungsten and Nitrogen Reacted Tungsten and Molybdenum Surfaces , 
Masters Thesis, Naval Postgraduate School, Monterey, California, 
December 1983. 

50. Webb, S.M., Study of Computer Simulation of Sputtering from 
Nitrogen Reacted Molybdenum and Tungsten Targets , Masters Thesis, 
Naval Postgraduate School, Monterey, California, June 1986. 

51. Clavenna, L.R. and Schmidt, L.D., "Interaction of N-2 with (100) 
W, " Surface Science , v. 22, pp. 365-391, 1970. 

52. Adams, D.L. and Germer , L.H., "The Adsorption of Nitrogen on 
W(100)," Surface Science , v. 26, pp. 109-124, 1971. 

53. Adams, D.L. and Germer, L.H., "Adsorption on Single Crystal Planes 
of Tungsten I. Nitrogen," Surface Science , v. 27, pp. 21-44, 1971. 

54. Griffiths, K., and others, "Adsor bate- I nduced Contracted Domain 
Structure: Nitrogen Chemisorbed on W {001 j," Phys i ca I Rev i ew Let- 
ters , v. 46, pp. 1584-1567, 1981. 

55. Winters, H.F. and Taglauer, E., "The Sputtering of Chemisorbed 
Nitrogen From Single Crystal Planes of Tungsten and Molybdenum," 
IBM Research Report , 1986. 

56. Winters, H.F., Schlaegel, J., and Home, D. , "RF Induction Tech- 
nique for Sample Heating in Surface Science Experiments," Jour na 1 
of Vacuum Science and Technology , v. 15, pp. 1605-1608, 1978. 



83 



57. JdKas, Mario M. and Harrison, D.E., Jr., "A Comparison Between 
Multiple Interaction Computer Simulations and the Linear Theory of 
Sputtering," Nuclear I n struments and Methods in Physics Research , 
v. B14, pp. 535-541, 1986. 

58. Webb, R.P., Harrison, D.E., Jr., and Jakas, M.M., "The Computer 
Simulation of Ion Induced Atomic Collision Cascades," Nuc I ear 
Instruments and Methods in Physics Research , v. B15, pp. 1-7, 1986. 

59. Harrison, D.E., Jr. and Jakas, M.M., "Simulation of the Atomic 
Collision Cascade," Radiation Effects , v. 99, pp. 153-169, 1986. 

60. Chemical Rubber Company, Handbook of Chemistry and Physics , 54th 
ed, CRC Press, 1973. 

61. Askland, D.R., The Science and Engineering of Materials , PWS Engi- 
neer i ng, 1984. 

62. Simons, Eric N. , Guide to the Uncommon Metals , Hart, 1967. 

63. General Electric, Chart of the Nuc l ides , 1975. 

64. Harrison, D.E., Jr. and Webb, R.P., "A Molecular Dynamics Simu- 
lation Study of the Influence of the Lattice Atom Potential Func- 
tion upon Atom Ejection Processes," Journal of Applied Physics , v, 
53, pp. 4193-4201, 1982. 

65. Huber, K.P. and Herzberg, G. , Constants of Diatomic Molecules , Van 
Nostrand Reinhold Company, 1979. 

66. Torrens, I.M., Interatomic Potentials , Academic Press, 1972. 

67. Bauer, E. and Poppa, H. , "The Interaction of Oxygen with the 
Mo (100) Surface," Surface Science, v. 88, pp. 31-64, 1979. 



84 



INITIAL DISTRIBUTION LI ST 



No. Copies 

1. Defense Technical Information Center 2 
Cameron Station 

Alexandria, Virginia 22304-6145 

2 . Super i ntendent 2 
Attn: Library, Code 0142 

Naval Postgraduate School 
Monterey, California 93943-5002 

3. Department Chairman, Code 61Sq 2 
Department of Physics 

Naval Postgraduate School 
Monterey, California 93943-5000 

4. Dr. Don E. Harrison, Jr. Code 61Hx 5 
Department of Physics 

Naval Postgraduate School 
Monterey, California 93943-5000 

5. Captain Philip J. Mattson, USA 6 
906 Park Avenue 

Manteca, California 95336 

6. Lieutenant Dana Majors, USN 1 
1064 Halsey Dr ive 

Monterey, California 93940 



85 



T | 



17 6 9 8 2 



NAVAL Pi H0QL 

MONTEREY, CALIFORNIA 93943-6008 



221101 



i 




chetra- 
from the 



L 



sotbed nitro«» * 

(100) P la " eS ° a compari- 

tfon ana -P«i- ntal 
results. 




Thesis 
M38345 

c.l 



221101 



Mattson 

Sputtering of chemi- 
sorbed nitrogen from tae 
(100) planes of tungstsn 
and molybdenum a comparx- 
son of computer simula- 
tion and experimental 
results .